Motorcycle gear shifting prompt evaluation method through intelligent algorithm

Abstract

The invention provides a motorcycle gear shifting prompt evaluation method through an intelligent algorithm, and the method comprises the steps: analyzing the deviation degree of a short response time ratio and a pre-judgment behavior standard threshold value, and distinguishing an active pre-judgment operation record and a passive response prompt operation record of a driver through the combination of motion connection interval distribution; performing aggregation statistics of response time distribution characteristics on the real response sub-record set, identifying a centralized trend position and a discrete fluctuation amplitude of a time interval in the real response sub-record set, and determining an optimization evaluation duration of a real prompt guide effect; and evaluating and optimizing the evaluation duration to shorten the improvement amplitude of the original response time reference value, obtaining a real prompt guide response validity score, evaluating the coincidence degree of the real prompt guide response behavior number proportion and the pre-judgment behavior standard threshold value, and obtaining accurate evaluation output of the motorcycle gear shifting prompt effect.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for evaluating motorcycle gear shift prompts using intelligent algorithms. Background Technology

[0002] In the field of intelligent modern transportation, motorcycle gear shift prompts directly impact driving safety and operational efficiency. This technology intelligently guides drivers to shift gears at the appropriate time, optimizing vehicle performance and reducing driver fatigue, making it an indispensable part of intelligent transportation development. With technological advancements, such systems are increasingly becoming crucial for enhancing the driving experience. However, existing methods for evaluating gear shift prompt effectiveness often overlook the complex impact of changes in driver behavior. Many evaluation methods focus only on surface data, such as reaction time, failing to analyze whether the driver is truly guided by the system. This approach is prone to misjudgment, especially after drivers develop personal habits through long-term use. Accurately distinguishing whether a driver's reaction stems from system prompts or their own pre-judgment is critical to prevent the system's true effectiveness from being obscured and to avoid discrepancies between evaluation results and actual effects. This approach focuses on identifying and judging "pre-response behavior," specifically whether the driver has already begun preparing to shift gears before the system prompts. Existing methods, failing to effectively capture and analyze the timing and motivation of this behavior, cannot distinguish whether a rapid reaction is the result of prompt guidance or an autonomous action based on the driver's experience. The confusion between these two factors leads to contradictions in the evaluation process, affecting the accurate judgment of the system's guidance effectiveness. Specifically, in real-world driving scenarios, drivers may have already placed their hands on the clutch or even begun shifting gears by observing changes in engine speed or vehicle speed before the system's prompt sound or signal appears. In this case, the system records a very short reaction time, seemingly indicating a good effect. However, the driver's actions are actually unrelated to the prompt and are based on a pre-judgment behavior based on personal experience. This misalignment in timing and confusion regarding behavioral motivation renders the evaluation results unreliable. Therefore, accurately identifying the actual reactions triggered by system prompts in complex driving behaviors and eliminating the influence of the driver's autonomous pre-judgment becomes a key issue in improving the evaluation of motorcycle shift prompt systems. Summary of the Invention

[0003] This invention provides a method for evaluating motorcycle gear shift prompts using intelligent algorithms, mainly including:

[0004] Acquire time interval records related to the driver's gear shifting operation. The time interval records include the time difference between the clutch pre-action moment and the system prompt moment, as well as its positive and negative distribution characteristics.

[0005] The predicted behavior recognition time window length and the predicted behavior standard threshold are determined based on the time interval records.

[0006] Determine the proportion of short response times, analyze the deviation of the proportion of short response times from the predicted behavior standard threshold, and distinguish between driver active prediction operation records and passive response prompt operation records;

[0007] The discrimination threshold is determined by the action connection interval distribution of the active prediction operation record and the passive response prompt operation record.

[0008] The discrimination threshold is adjusted based on the individual differences actively predicted by the driver, and the time interval records are filtered to obtain a set of real response sub-records;

[0009] The response time distribution characteristics of the real response sub-record set are statistically analyzed to determine the optimized evaluation time, and the improvement of the optimized evaluation time on the original response time is evaluated. The evaluation result of the shift prompt effect is then output.

[0010] Furthermore, the acquisition of time interval records related to the driver's gear shifting operations includes:

[0011] The clutch pedal pressure sensor collects the moment when the driver begins to apply force with his foot as the clutch preparation moment.

[0012] Record the time when the vehicle's gear shift indicator emits an audible and visual signal.

[0013] Calculate the time difference between the clutch preparatory action moment and the prompt issuance moment, and generate the time interval record containing positive and negative distribution characteristics.

[0014] Furthermore, after acquiring the time interval records related to the driver's gear shifting operation, based on the continuously collected gear shifting operation time difference sequence in the time interval records, the ratio of the number of occurrences of negative difference values ​​to the total number of occurrences is calculated, and the time difference sequence is segmented using the sliding time window method to determine the length value of the prediction behavior recognition time window.

[0015] The median of the positive differences is extracted from the time interval records and used as the original response time baseline value.

[0016] Furthermore, the step of determining the predictive behavior standard threshold based on the time interval records includes:

[0017] The standard deviation of the time difference within the negative value interval is calculated by using the ratio of the number of occurrences of negative differences in the time interval records to the total number of occurrences.

[0018] The data density is determined based on the ratio of the standard deviation to the length of the predicted behavior recognition time window.

[0019] The data density is compared with a preset threshold to generate the predicted behavior standard threshold.

[0020] Furthermore, determining the proportion of short response times includes: calculating the proportion of negative data points to the total number of records by using the negative time difference data recorded in the time interval as the proportion of short response times.

[0021] Furthermore, after determining the proportion of short response times, the process includes:

[0022] Calculate the difference between the proportion of short response times and the predicted behavior standard threshold;

[0023] The direction of deviation of the driver's anticipated behavior is determined based on the sign of the difference;

[0024] If the difference is positive and exceeds the preset deviation threshold, then continuous negative data points are extracted from the negative value interval recorded in the time interval, and the number of data points within the unit time window is counted as the data density value to obtain the density distribution reflecting the concentration of the predicted behavior.

[0025] Based on the location of the peak region in the density distribution, the distribution ratio of the short response time in different negative sub-intervals is extracted. The ratio of the number of data points in each sub-interval to the total number of negative data points is calculated as the contribution weight. The timing advance of the predicted behavior is obtained by weighted summation of the distribution ratio of each negative sub-interval using the contribution weight.

[0026] Furthermore, determining the discrimination threshold based on the action connection interval distribution of the active prediction operation record and the passive response prompt operation record includes:

[0027] Calculate the ratio of the time lead to the length of the predicted behavior recognition time window;

[0028] Using the aforementioned proportional coefficient as a boundary standard, operation records with a value less than the product of the proportional coefficient and a preset benchmark value are selected from the action connection interval sequence.

[0029] The selected operation records are classified according to the sign of the time interval value. Records with negative time interval values ​​are marked as active prediction operation records, and records with positive time interval values ​​are marked as passive response prompt operation records.

[0030] By calculating the average action connection interval of the active prediction operation record and the average action connection interval of the passive response prompt operation record, typical characteristic values ​​of the two types of operations are determined.

[0031] Based on the midpoint of the two means and the standard deviation of the action connection interval distribution, a discrimination threshold is generated to distinguish between the two types of operation records.

[0032] Furthermore, after determining the discrimination threshold, the process includes:

[0033] The change in capacitance of the shift lever touch sensor is detected within the data segment prior to the time of system prompt issuance, filtered from the time interval record.

[0034] If the capacitance value exceeds the preset touch threshold, the moment when the driver's hand touches the gear shift lever is recorded, and the time difference between the moment the gear shift lever is touched and the moment the system prompts the system is activated is calculated.

[0035] If the time difference is negative and its absolute value exceeds a preset threshold, it is determined that the hand pre-touching has been completed.

[0036] By using the hand pre-touch completion state, the clutch pedal pressure curve and shift lever position signal for the corresponding time period are extracted from the action connection interval data to determine the sequence of clutch disengagement and shift lever positioning.

[0037] If the preset coordination threshold is met, the driver's active gear shifting operation mode is determined by combining road slope data and real-time vehicle speed values.

[0038] Furthermore, the step of adjusting the discrimination threshold of the proactive prediction operation record based on individual differences in the driver's proactive prediction, and filtering the time interval records to obtain the set of true response sub-records, includes:

[0039] Obtain the intensity values ​​of early gear shifting habits of different drivers, and calculate the difference ratio between each driver's habit intensity value and the average habit intensity value of all drivers;

[0040] The individual difference correction factor is determined based on the distribution range of the difference ratio;

[0041] Calculate the ratio of the predicted behavior advance amount to the length of the predicted behavior recognition time window;

[0042] The correction magnitude is determined by multiplying the ratio by the individual difference correction factor, and the upper and lower boundary values ​​of the discrimination threshold of the active prediction operation record are adjusted accordingly.

[0043] The time interval records are filtered one by one according to the adjusted boundary, and the records located between the lower boundary value and the upper boundary value are retained.

[0044] Check whether the shift operation corresponding to the filtered record contains a complete process. If the condition is met, include it in the set of real response sub-records.

[0045] Furthermore, the step of performing response time distribution characteristic statistics on the set of real response sub-records to determine the optimized evaluation duration, and evaluating the improvement of the optimized evaluation duration on the original response time, includes:

[0046] Sort the time interval data in the set of real response sub-records, and calculate the median and mean of the time interval data;

[0047] The symmetry of the data distribution is determined by comparing the difference between the median and the mean, and the location of central tendency is analyzed.

[0048] Based on the central tendency position, calculate the deviation of each time interval data from the central tendency position, and obtain the standard deviation to measure the discrete fluctuation amplitude;

[0049] The optimization evaluation duration is determined by removing outliers and recalculating the average value based on the central trend position and the discrete fluctuation amplitude;

[0050] The extent of improvement is assessed based on the difference between the optimized assessment time and the original response time baseline.

[0051] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0052] This invention discloses a motorcycle gear shift prompt evaluation method using intelligent algorithms. It addresses the problems of traditional evaluation methods, such as difficulty in distinguishing between the driver's proactive anticipation and passive response prompts, and the neglect of individual early gear shifting habits leading to evaluation bias. The method involves real-time acquisition of the driver's hand-clutch preparation action, gear shift lever contact moment, and the prompt issuance moment, measuring the time intervals and their distribution characteristics. Simultaneously, it obtains the anticipation behavior recognition time window length, anticipation behavior standard threshold, and original response time benchmark value. The method evaluates the correspondence between the time interval record and the window length, calculates the proportion of short response times to reflect the intensity of early gear shifting habits, extracts the action connection interval, and analyzes the deviation of the proportion from the threshold. This method extracts data density and contribution weights from negative value intervals, identifies the proportion of anticipatory behavior advance, and distinguishes between active and passive records based on interval distribution. It then obtains the completion status of the hand touching the gear shift lever before the system prompt, the clutch disengagement and gear shifting sequence, determines the active operation mode, adjusts the effectiveness judgment time limit using habit intensity as a correction factor, filters the set of real response sub-records, performs distribution feature aggregation statistics, determines the optimized evaluation time, evaluates its improvement to the baseline value, obtains an effectiveness score, and integrates the quantity ratio, correction magnitude, habit intensity, and active record ratio to generate a comprehensive evaluation report. Finally, the evaluation report assesses the degree of conformity between the real response ratio and the threshold, outputting accurate evaluation results. This method effectively improves the evaluation accuracy of motorcycle gear shift prompt guidance effect, optimizing driving safety and experience. Attached Figure Description

[0053] Figure 1 This is a flowchart of a motorcycle gear shift prompt evaluation method based on an intelligent algorithm according to the present invention.

[0054] Figure 2 This is a schematic diagram of a motorcycle gear shift prompt evaluation method based on an intelligent algorithm according to the present invention.

[0055] Figure 3 This is another schematic diagram of a motorcycle shift prompt evaluation method based on an intelligent algorithm according to the present invention. Detailed Implementation

[0056] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0057] like Figure 1-3 This embodiment of a motorcycle gear shift prompt evaluation method using intelligent algorithms may specifically include:

[0058] Step S101: Real-time acquisition of the driver's hand clutch preparation action time, gear shift lever contact time, and system prompt issuance time. Combining the time interval value between the preparation action time and the prompt issuance time and its positive and negative distribution characteristics, the time interval record is obtained. At the same time, the predicted behavior recognition time window length value, the predicted behavior standard threshold, and the original response time benchmark value are obtained.

[0059] The timing of the driver's initial foot pressure application is collected from a pressure sensor mounted on the motorcycle clutch pedal as the clutch preparation moment. The initial hand contact with the shift lever is detected by a touch sensor on the shift lever base. The timing of the onboard shift warning device's audible and visual signal is recorded. The time difference between the clutch preparation moment and the warning signal timing is calculated. A negative time difference indicates that the driver has already begun preparing before the warning; a positive time difference indicates that the driver responded after the warning. This yields a time interval record containing positive and negative distribution characteristics. Based on the continuously collected shift operation time difference sequence in the time interval record, the ratio of the number of negative differences to the total number of differences is calculated. The time difference sequence is segmented using a sliding time window method to determine the length of the predicted behavior recognition time window. The median of the positive differences is extracted from the time interval record as the original response time reference value. The standard deviation of the time difference within the negative value interval is calculated by the ratio of the number of occurrences of the negative difference to the total number of occurrences. The data density is determined by the ratio of the standard deviation to the length of the predicted behavior recognition time window. If the data density exceeds a preset threshold, it is judged as a high prediction tendency, and the predicted behavior standard threshold is obtained.

[0060] Optionally, the sliding time window method is implemented by determining the window size based on the average frequency of gear shifting operations, such as a window every 5 seconds, and segmenting according to the continuity of the operation sequence to ensure that each window contains a complete operation cycle. The formula for calculating data density is density equal to the standard deviation divided by the time window length, with a threshold set such as 0.8. If the density is greater than 0.8, it is judged as having a high predictive tendency.

[0061] Specifically, the motorcycle shift indicator evaluation device monitors the driver's foot movements in real time by installing a piezoelectric pressure sensor under the clutch pedal. When the driver's foot begins to apply pressure to the clutch pedal, the pressure sensor detects the pressure change and records this moment as the clutch preparation moment. The trigger threshold of the pressure sensor is set to 10% of the total pedal travel pressure value. When the detected pressure value exceeds this threshold, the system determines that the driver has begun to prepare to shift gears.

[0062] For example, the gear shift lever base is equipped with a capacitive touch sensor that detects the contact state between the driver's hand and the gear shift lever. When the driver's palm or fingers touch the surface of the gear shift lever, the sensor determines the contact moment by detecting changes in capacitance. The onboard gear shift indicator is connected to the engine control unit via a CAN bus and triggers an audible and visual warning signal based on engine speed and vehicle speed information. The system accurately records the moment the warning signal is issued. The generation of the time interval record involves the precise acquisition and calculation of three key moments. The time difference between the clutch preparation moment and the warning issuance moment is calculated with millisecond precision. A negative difference indicates that the driver has already begun gear shift preparation based on their own experience before receiving the system warning. This pre-judgment behavior will affect the accuracy of the assessment of the actual guidance effect of the system.

[0063] In one possible implementation, the system uses a sliding time window method to process continuous gear shifting operation data. The length of the sliding window is dynamically adjusted according to the motorcycle model and driving environment; the window length is set shorter for urban driving and appropriately longer for highway driving. By statistically analyzing the frequency of negative differences within each window, the system can quantify the strength of the driver's predictive tendency.

[0064] Preferably, the standard threshold for predictive behavior is determined using statistical methods, by calculating the standard deviation of the time difference within the negative value interval to reflect the dispersion of the data. When the ratio of the standard deviation to the length of the predictive behavior recognition time window is lower than the preset threshold, it indicates that the driver's predictive behavior has high consistency and regularity, and thus it is determined that the driver has formed a stable gear shifting predictive habit.

[0065] Step S102: Evaluate the correspondence between the time interval records and the length of the predicted behavior recognition time window, take the proportion of negative time intervals in the time interval records as the proportion of short response time, evaluate the strength of the driver's early gear shifting habit formed based on historical driving experience, and extract the action connection interval between the gear shift lever contact time and the preparatory action time.

[0066] All negative time difference data are filtered from the time interval records. The proportion of negative data points to the total number of records is calculated as the short response time ratio. This short response time ratio is compared with the length of the predicted behavior recognition time window. When the short response time ratio exceeds a preset threshold, it is determined that the driver has a tendency to predict behavior, thus obtaining the frequency distribution characteristics reflecting the driver's autonomous initiation of gear shift preparation before system prompts. By analyzing the concentration and distribution range of negative time differences in the frequency distribution characteristics, the peak frequency of time differences occurring within the negative interval is statistically analyzed. If the peak frequency is located at a specific position within the negative interval and occurs more than a preset threshold, it is determined that the driver has formed a stable habit of shifting gears in advance. Optionally, when statistically analyzing the peak frequency of negative time differences, a histogram analysis method is used to divide the negative interval into several equal-width sub-intervals. The frequency of data points occurring in each sub-interval is calculated, and the sub-interval with the highest frequency is the peak frequency position. The concentration is determined by calculating the frequency ratio of the peak sub-interval and its adjacent sub-intervals, and the distribution range is characterized by the standard deviation of the negative interval. If the frequency peak occurs more than a preset threshold multiple times and is located at a specific position in the negative range, it is determined to be a stable early shifting habit. The intensity value of the early shifting habit is obtained based on the height and position of the frequency peak. The time interval records are then classified according to this intensity value. The time difference sequence between the shift lever contact moment and the clutch preparation moment is extracted from the classified records. The mean of this time difference sequence is calculated as the baseline value for the action connection interval. The variance is used to determine the fluctuation range of the action connection interval, thus obtaining the action connection interval distribution.

[0067] Specifically, the driver's predictive behavior characteristics are assessed through statistical analysis of time interval records. All negative time differences are filtered from continuously collected gear shifting operation data; these negative values ​​indicate that the driver began preparing to shift gears before the system prompts. The proportion of negative data points to the total number of records is the short response time percentage, which directly reflects the frequency of the driver's predictive behavior. There is a specific correlation between the short response time percentage and the length of the predictive behavior identification time window. When the short response time percentage reaches 60% or more, it indicates that the driver exhibits a tendency to predict gears in most shifting operations. By comparing these two parameters, it is determined whether the driver has a clear pattern of predictive behavior; when the percentage exceeds a preset threshold, the driver is considered to have a tendency towards predictive behavior.

[0068] For example, obtaining frequency distribution characteristics involves statistically analyzing the distribution density of negative time differences across different time intervals. The negative intervals are divided into multiple sub-intervals, and the number of data points in each sub-interval is counted to form a frequency histogram. By analyzing the peak position and shape of the histogram, the system can identify typical time characteristics of the driver's predictive behavior. If the frequency peak consistently appears in a specific negative interval, and the peak height exceeds a threshold, it is determined that the driver has formed a stable habit of shifting gears in advance.

[0069] Preferably, the calculation of the early shifting habit intensity value is based on two dimensions: the height and position of the frequency peak. The higher the peak, the more concentrated the driver's anticipatory behavior at that time point; the closer the peak position is to the far end of the negative value range, the earlier the driver's anticipatory time. The system integrates these two factors, with preset coefficients, and obtains a quantified habit intensity value through weighted calculation. This value is used for subsequent classification processing of time interval records.

[0070] In one possible implementation, the action transition interval is extracted by calculating the time difference between the gear shift lever engagement moment and the clutch preparation moment. These time difference values ​​are extracted from the categorized records to form a time difference sequence, and the stability of the action transition is evaluated by calculating the statistical characteristics of the sequence. The mean reflects the typical action transition time of the driver, while the variance reflects the consistency of action execution.

[0071] Step S103: Analyze the deviation between the proportion of short response time and the standard threshold for predicted behavior, and distinguish between driver's active prediction operation records and passive response prompt operation records by combining the distribution of action connection intervals.

[0072] The difference between the proportion of short response times and the standard threshold for predicted behavior is calculated. The direction of deviation of the driver's predicted behavior is determined based on the sign of the difference. Optionally, the standard threshold for predicted behavior is determined based on statistical analysis of historical driving data, taking the average of the proportion of short response times plus a standard deviation as the threshold. If the difference is positive and exceeds the preset deviation threshold, consecutive negative data points are extracted from the negative value interval recorded in the time interval. The number of data points within a unit time window is counted as the data density value of the negative value interval, obtaining a density distribution reflecting the concentration of predicted behavior. Based on the location of the peak region in the density distribution, the distribution ratio of the proportion of short response times in different negative sub-intervals is extracted. The ratio of the number of data points in each sub-interval to the total number of negative data points is calculated as the contribution weight of that sub-interval. The time advance of the predicted behavior is obtained by weighted summation of the contribution weights, and the proportional coefficient of the time advance to the length of the predicted behavior recognition time window is calculated. Using the aforementioned proportional coefficient as a dividing standard, operation records with intervals smaller than the product of the proportional coefficient and a preset benchmark value are selected from the action connection interval sequence. The selected operation records are then classified based on the sign of their corresponding time interval values: records with negative time intervals are marked as proactive prediction operation records, and records with positive time intervals are marked as passive response prompt operation records. Optionally, the preset benchmark value is set to 0.5. Multiple experiments are conducted to ensure the rationality and operability of the selection standard. By calculating the mean action connection intervals of the proactive prediction operation records and the passive response prompt operation records, typical characteristic values ​​for the two types of operations are determined. Based on the midpoint of the two means combined with the standard deviation of the action connection interval distribution, a discrimination threshold for distinguishing between proactive prediction operation records and passive response prompt operation records is obtained.

[0073] Specifically, the degree of a driver's predictive tendency is assessed by calculating the deviation between the proportion of short response times and a predictive behavior standard threshold. The predictive behavior standard threshold is typically set as the average proportion of short response times in historical data under normal response patterns. When the actual measured proportion of short response times exceeds this threshold, it indicates a significant predictive tendency in the driver. In another approach, the deviation calculation involves two dimensions: the absolute value of the difference and the relative deviation rate. The data density in the negative value interval reflects the distribution characteristics of the driver's predictive behavior over time. The negative value interval is divided into multiple consecutive time windows, each 100 milliseconds wide, and the number of data points within each window is counted. The data density value is calculated as the ratio of the number of data points per unit time window to the length of that window. When the data density value of a certain time window exceeds a preset threshold, that time period is determined to be a high-incidence interval for predictive behavior. This density distribution map can visually display the temporal patterns of the driver's predictive behavior, helping to identify typical predictive patterns.

[0074] For example, the contribution weights are calculated using a weighted method based on data distribution. First, peak regions in the density distribution map are identified; these regions represent the times when drivers most frequently make predictions. For each negative sub-interval, the proportion of data points within that sub-interval to the total number of negative data points is calculated; this proportion is the initial weight for that sub-interval. Subsequently, the initial weights are adjusted based on the sub-interval's distance from zero; the farther the sub-interval is from zero, the more pronounced its prediction characteristics, and the larger the corresponding weight adjustment coefficient. The overall prediction advance time is obtained by multiplying the adjusted weights of each sub-interval by their corresponding time values ​​and summing the results. Determining the proportional coefficient involves evaluating the relationship between the prediction advance time and the length of the prediction behavior recognition time window. The prediction behavior recognition time window length is a preset reference benchmark, typically set between 1.5 and 2.5 seconds depending on the motorcycle type and road conditions. When the prediction advance time exceeds 40% of the window length, the system considers the driver to have a strong prediction tendency.

[0075] In one possible implementation, this scaling factor is used not only to assess the predicted intensity.

[0076] Preferably, the boundary standard for the action connection interval is determined by multiplying a proportional coefficient by a preset benchmark value. The preset benchmark value is typically taken from the average action connection interval in the normal response mode of historical data, approximately 300 to 500 milliseconds. When the action connection interval of an operation record is less than the product of the proportional coefficient and the benchmark value, it is classified into the fast response category. The classification is further based on the sign of the corresponding time interval value: records with negative time intervals and belonging to fast response are marked as proactive prediction operations, while records with positive time intervals but slower responses are marked as passive response operations.

[0077] Specifically, the proactive prediction operation records exhibit a particular behavioral pattern. In these records, the driver often begins preparing to perform the action before the engine speed approaches the shift point, the clutch pedal pressure change curve shows a gradual increase, and the gear shift lever engagement is more decisive. Analysis of the action transition interval distribution in these records reveals that its mean is typically around 200 milliseconds, with a small standard deviation, indicating a high degree of consistency in action execution.

[0078] Understandably, passive response prompt operation records exhibit different characteristics. In these records, the driver only begins to act after hearing or seeing the system prompt, resulting in a relatively long reaction time, with the average action transition interval typically exceeding 600 milliseconds. More importantly, the action transition interval for these operations fluctuates significantly, with a standard deviation markedly higher than that for active anticipation operations, reflecting relatively poor driver coordination during passive responses.

[0079] For example, the discrimination threshold is determined based on the statistical differences between the two types of operation records. The mean action transition intervals for proactive prediction operation records and passive response operation records are calculated separately, and the midpoint of the two means is taken as the initial discrimination point. Considering the distribution characteristics of the actual data, further adjustments are made based on the standard deviation of the action transition interval distribution. When the distributions of the two types of records overlap, Bayesian decision theory is used to determine the optimal discrimination threshold, minimizing the classification error rate.

[0080] In one embodiment, a time series analysis method is also introduced to improve classification accuracy. By analyzing the changing trends of time intervals between multiple consecutive gear shifts, the evolution of the driver's anticipatory behavior can be identified. When three or more consecutive operations exhibit anticipatory characteristics, the weight of the driver's subsequent anticipatory tendencies is increased, thereby achieving a dynamically adaptive classification strategy. Through the above multi-dimensional analysis and discrimination mechanism, the driver's active anticipatory operations and passive response operations can be accurately distinguished, providing reliable data support for evaluating the actual effect of the gear shift prompt system. This discrimination mechanism takes into account the influence of individual driver differences and driving habits, avoiding the misjudgment problems that may arise from simply judging based on reaction time.

[0081] The system records the completion status of the driver's hand touching the gear shift lever before the system prompt is issued by the time interval. The system also collects the arrival status of the clutch pedal being pressed to the disengagement point and the positioning status of the gear shift lever in the target gear from the action connection interval. The system analyzes the operation completion status of the driver touching the gear shift lever before the system prompt, evaluates the timing sequence of clutch disengagement and gear shift lever positioning, and determines the operation mode of the driver actively initiating the gear shift process based on road conditions and vehicle speed.

[0082] The system filters data segments from the time interval records prior to the system prompt issuance time, detects changes in capacitance values ​​of the shift lever touch sensor during this period. If the capacitance value exceeds a preset touch threshold, the moment the driver's hand touches the shift lever is recorded. The time difference between this touch moment and the system prompt issuance time is calculated. When the time difference is negative and its absolute value exceeds the preset threshold, it is determined that the hand pre-touching is complete. Based on the hand pre-touching completion state, the clutch pedal pressure curve for the corresponding time period is extracted from the action connection interval data. The moment when the pressure value reaches the pressure value required for complete clutch disengagement is detected as the disengagement arrival moment. Simultaneously, the gear position change signal is obtained from the shift lever position sensor, and the moment the shift lever moves from neutral to the target gear is recorded. The time difference between the disengagement arrival moment and the arrival moment is calculated. The order of clutch disengagement and shift lever arrival is determined based on the time difference. If clutch disengagement occurs earlier than shift lever arrival and the time difference is less than a preset coordination threshold, the driver's operation is determined to have good timing coordination characteristics. Combining the hand pre-touching completion state determination result and timing coordination characteristics, the strength of the driver's proactive anticipation of gear shifting behavior is evaluated. Using the aforementioned behavioral characteristic intensity value, current road condition slope data and real-time vehicle speed values ​​are obtained from on-board sensors. When the combination of slope data and vehicle speed values ​​meets the preset shifting conditions and the behavioral characteristic intensity value exceeds the threshold, the driver is determined to actively initiate the shifting process based on road conditions and vehicle speed.

[0083] Specifically, in one implementation, the driver's pre-touch behavior is identified by analyzing the data segment in the time interval record prior to the issuance of the system prompt.

[0084] Specifically, the capacitive touch sensor mounted on the gear shift lever base can detect minute changes in capacitance. When the driver's hand approaches the gear shift lever, even without fully gripping it, the sensor can capture the trend of capacitance change. Multiple touch thresholds are set: the first threshold corresponds to the state of approaching but not touching, the second to a light touch, and the third to a full grip. By detecting when the capacitance reaches different thresholds, the system can accurately record the contact process between the driver's hand and the gear shift lever. The determination of the pre-touch and completed state considers not only the time difference between the touch moment and the system prompt moment but also the continuity and stability of the touch. When the time difference is negative and the absolute value exceeds 500 milliseconds, it indicates that the driver has already begun preparing to shift gears before the system prompt. Further analysis of the capacitance change curve reveals that if the curve shows a steady rise and remains high, it is determined to be a stable pre-touch and completed state; if the curve fluctuates or falls back after a brief touch, it is determined to be a tentative touch and is not counted as a pre-touch and completed state. This multi-dimensional determination method can accurately identify the driver's true operating intention and avoid misjudgments caused by accidental contact.

[0085] For example, detecting the clutch disengagement arrival time involves precise analysis of the pedal pressure curve. A pressure sensor mounted under the clutch pedal collects pressure data at a high frequency, forming a continuous pressure variation curve. Critical moment points are identified by analyzing the slope changes of the curve. The initial disengagement moment is recorded when the pressure value reaches the critical value for clutch disengagement to begin, and the disengagement arrival moment is recorded when the pressure value reaches the pressure value required for complete disengagement. The complete disengagement pressure value is typically 70% to 80% of the maximum pedal pressure; this range is adjusted according to the clutch characteristics of different motorcycle models. The smoothness of the pressure curve is also monitored. If a pressure drop or fluctuation occurs during disengagement, it indicates insufficient driver skill; this information is also recorded for subsequent analysis.

[0086] In one possible implementation, the shift lever's positioning status is detected using a combination of a position sensor and an angle sensor. The position sensor detects the shift lever's position in each gear, while the angle sensor measures the shift lever's rotation angle. The starting moment is recorded when the shift lever begins to move from neutral; the positioning moment is recorded when the shift lever reaches the target gear and remains stably in place for more than 100 milliseconds. The smoothness of the operation is assessed by analyzing the shift lever's movement trajectory. A direct and rapid trajectory indicates driver proficiency; multiple adjustments or pauses indicate hesitation in the operation.

[0087] Preferably, the evaluation of timing coordination characteristics is based on the time relationship between clutch disengagement and shift lever positioning. In an ideal shifting operation, the clutch should be fully disengaged before the shift lever begins to move and released appropriately after the shift is completed. The time difference between the disengagement arrival time and the positioning time is calculated. When the clutch disengagement is 50 to 200 milliseconds earlier than the shift lever positioning, it is considered good timing coordination. If the time difference is too small or negative, it indicates that the clutch disengagement is insufficient, which may lead to shifting difficulties. If the time difference is too large, it indicates that the clutch disengagement time is too long, which may increase wear. In one approach, the calculation of the behavioral characteristic intensity value comprehensively considers multiple dimensions of operational characteristics. The system assigns weights to various indicators such as hand pre-touch completion state, timing coordination characteristics, and operational smoothness. After normalization of each indicator, a comprehensive behavioral characteristic intensity value is obtained by weighted summation. The pre-touch completion state accounts for 40% of the weight, timing coordination characteristics account for 35%, and operational smoothness accounts for 25%. When the behavioral characteristic intensity value exceeds a preset threshold, the system determines that the driver has the ability to actively anticipate shifting.

[0088] Understandably, the collection and analysis of road condition and vehicle speed data are crucial for determining the active start mode. Onboard accelerometers and gyroscopes detect road gradient, obtaining stable gradient data through continuous sampling and filtering. Vehicle speed sensors monitor the current speed in real time and calculate the rate of change of speed. A matrix of shifting conditions is preset for different combinations of road conditions and vehicle speeds; for example, shifting at lower RPMs to maintain power when going uphill, and shifting at higher RPMs to utilize engine braking when going downhill.

[0089] For example, when an uphill slope exceeding 10 degrees and a vehicle speed below 30 kilometers per hour is detected, it is determined that a downshift is needed to obtain greater torque. If the driver's behavioral characteristic intensity value exceeds a threshold at this time, and the driver begins shifting preparation before the system prompts them, it is determined that the driver actively initiated the shifting process based on road conditions and vehicle speed. This proactive initiation mode indicates that the driver has good driving experience and predictive ability, and can make the correct shifting decision in advance based on actual road conditions.

[0090] In one embodiment, the system also records and analyzes changes in the driver's operating patterns in different scenarios. Through long-term data accumulation, the system can identify the evolution of the driver's operating habits, from initial passive response to proactive prediction. This process reflects the improvement of driving skills and increased familiarity with vehicle characteristics.

[0091] Step S104: Using individual differences in the intensity of early gear shifting habits as a correction factor, the proportion of the predicted behavior's advance amount to the length of the predicted behavior recognition time window is used as the boundary for adjusting the validity judgment time limit. The corrected validity judgment time limit is generated and applied to filter time interval records, from which a set of real response sub-records that meet the corrected time limit is extracted. Optionally, the formula for calculating the correction magnitude is: Correction Magnitude = Proportion × Correction Factor Value, where the proportion is the percentage of the predicted behavior's advance amount to the length of the predicted behavior recognition time window, and the correction factor value is the difference ratio multiplied by a preset coefficient. The multiplication method is used because the proportion reflects individual behavioral characteristics, and the correction factor reflects the degree of difference; combining the two allows for a comprehensive adjustment of the magnitude. For example, if the proportion is 20% and the correction factor is 2, then the correction magnitude is 40%. If the original validity judgment time limit is 10 seconds, then the upper boundary value is 10 plus 40%, which is 14 seconds, and the lower boundary value is 10 minus 40%, which is 6 seconds.

[0092] The intensity values ​​of early gear shifting habits of different drivers are obtained. The difference ratio between each driver's habit intensity value and the average habit intensity value of all drivers is calculated. An individual difference correction factor is determined based on the distribution range of the difference ratio. If the difference ratio exceeds a preset threshold, the difference ratio is multiplied by a preset coefficient to obtain the correction factor value. Using the correction factor value, the proportion of the predicted behavior advance amount to the predicted behavior recognition time window length is calculated. The correction magnitude is determined by multiplying the proportion by the correction factor value. The original validity judgment time limit is adjusted using the correction magnitude, where the upper boundary value is the original limit plus the correction magnitude, and the lower boundary value is the original limit minus the correction magnitude, resulting in the corrected validity judgment time limit. Based on the corrected validity judgment time limit, time interval records are filtered one by one. If the time interval value of a record is between the lower boundary value and the upper boundary value, the record is retained, and records outside the limit range are removed, resulting in a preliminary set of records that meet the conditions. Using the preliminarily qualified record set, check whether the gear shifting operation corresponding to each record includes a complete clutch disengagement, gear shifting, and clutch release process. If the gear shifting process is complete and the time interval between records is continuous, then the record is included in the real response sub-record set.

[0093] Specifically, in one implementation, an individual difference assessment mechanism is established by analyzing the distribution of the intensity of drivers' pre-shifting habits. Each driver's pre-shifting habit intensity value is obtained; this value reflects the driver's tendency to proactively prepare for shifting gears before system prompts, with higher values ​​indicating more pronounced anticipatory behavior. By calculating the arithmetic mean of all drivers' habit intensities, the system establishes a group behavior benchmark. The difference ratio between each driver's habit intensity value and this average is the quantitative representation of individual differences. The calculation of the difference ratio involves the concept of relative deviation. When a driver's habit intensity value is 1.8, while the group average is 1.2, the difference ratio is 0.5, indicating that the driver's anticipatory tendency is 50% higher than the average. The system sets a threshold range for the difference ratio, typically between ±0.3. When the difference ratio exceeds this range, it indicates a significant difference between the driver's operating habits and the group, requiring personalized correction. The correction factor is determined using a linear mapping method; the difference ratio is multiplied by a preset coefficient of 0.6, and the resulting value is the driver's individual difference correction factor.

[0094] For example, the calculation of the ratio of the anticipation advance to the time window length involves the analysis of the relationship between two key parameters. The anticipation advance refers to the time length before the driver begins the gear shifting action before the system prompts, typically between 200 and 800 milliseconds. The anticipation recognition time window length is the time range used by the system to detect the anticipation, generally set to 2000 milliseconds. When the anticipation advance is 600 milliseconds and the window length is 2000 milliseconds, the ratio is 600 / 2000. This ratio reflects the proportion of the driver's anticipation within the entire recognition window and is an important indicator for evaluating the strength of the anticipation. Multiplying this ratio by a correction factor yields the correction magnitude, used to adjust the validity judgment time limit.

[0095] In one possible implementation, the adjustment of the validity discrimination time limit employs a two-way expansion mechanism. The original validity discrimination time limit is typically set between 300 and 1000 milliseconds, a range derived from statistical analysis of a large amount of normal response data. When the correction increment is 150 milliseconds, the upper boundary of the original limit is increased by 150 milliseconds, becoming 1150 milliseconds; the lower boundary is decreased by 150 milliseconds, becoming 150 milliseconds. This two-way adjustment method can simultaneously accommodate both overly fast and slow reactions, allowing the limit range to better adapt to individual differences. The adjusted limit range is dynamically updated as driver operating habits change, achieving a personalized evaluation standard.

[0096] Preferably, the time interval recording screening process employs a multi-level filtering mechanism. First, all time interval records undergo preliminary screening, comparing the time interval values ​​with corrected boundary ranges. Records falling between the upper and lower boundaries are retained, while those exceeding the range are marked as abnormal data. For the retained records, their temporal continuity is further checked; the time interval between adjacent records should not exceed 5 seconds; otherwise, they are considered independent operation sequences. Through this continuity check, valid recording sequences belonging to the same driving process can be identified.

[0097] Specifically, determining the completeness of a gear shift operation involves detecting three key stages. The clutch disengagement stage requires the clutch pedal pressure to reach the disengagement threshold and remain stable; the gear shifting stage requires the shift lever to move from the initial gear to the target gear and remain stably in place; and the clutch release stage requires the pedal pressure to gradually return to its initial state. The system verifies through sensor data whether these three stages are present and executed in the correct order. Only records containing all three stages are considered valid gear shift operation records.

[0098] Understandably, assessing the continuity between records considers not only the time dimension but also the consistency of operating patterns. Analyzing the operational characteristics of adjacent records includes parameters such as clutch disengagement depth, shift speed, and action transition time. When multiple consecutive records exhibit similar operating patterns, it indicates that the driver's operation is stable and repeatable. This consistency is a crucial criterion for determining whether a record represents a genuine response.

[0099] For example, in congested urban traffic, drivers frequently perform low-speed gear shifts. If the system detects that in 8 out of 10 consecutive gear shifts by a driver, the time interval falls within the corrected limits, and each shift includes all three stages, and the time intervals between these records are all within 3 seconds, the operation pattern shows a high degree of consistency, then the system will include all 8 records in the real response sub-record set as valid data for evaluating the driver's gear shift prompt response effectiveness.

[0100] In one embodiment, a confidence score mechanism is also introduced to enhance the accuracy of the screening. For each record that passes the initial screening, a confidence score is calculated based on factors such as its distance from the corrected boundary center value, the degree of operational integrity, and consistency with adjacent records. Only records with a confidence score exceeding 0.7 are ultimately included in the set of true response sub-records.

[0101] Step S105: Perform aggregated statistics on the response time distribution characteristics of the real response sub-record set, identify the central tendency position and discrete fluctuation amplitude of the time interval within the real response sub-record set, and determine the optimized evaluation duration of the real prompt guidance effect.

[0102] All time interval data in the real response sub-record set are sorted, and the median and mean of the time interval data are calculated. The symmetry of the data distribution is determined by comparing the difference between the median and the mean. If the difference exceeds a preset threshold, the median is used as the central tendency position; if the difference is within the threshold, the mean is used as the central tendency position. Based on the central tendency position, the deviation of each time interval data from the central tendency position is calculated. The variance is obtained by summing the squares of the deviations and dividing by the number of records. The standard deviation is obtained by taking the square root of the variance. The standard deviation is used to measure the dispersion of the time interval data. Based on the central tendency position and the dispersion, outliers exceeding the range of the central tendency position plus or minus two standard deviations are removed. The average of the data after removing outliers is recalculated, and the recalculated average is used as the optimized evaluation duration of the real prompt guidance effect.

[0103] Specifically, in one implementation, a comprehensive statistical analysis is performed on the set of real response sub-records to reveal the distribution patterns of driver response times. First, all time interval data in the set are arranged in ascending order, forming an ordered sequence. This sorting process allows for a direct observation of the data's distribution range and density. Comparing the median and mean is an important method for determining the symmetry of the data distribution. When the data exhibits a normal distribution, the difference between the median and the mean is small, typically within 5%. However, when extreme values ​​exist or the data distribution is skewed, the mean is pulled towards the extremes, while the median remains relatively stable. Setting a difference threshold of 10% of the mean, if the difference between the median and the mean exceeds this threshold, it indicates a significant skewness in the data distribution. In this case, choosing the median as the location of central tendency better represents the central characteristics of the data.

[0104] For example, the calculation of discrete fluctuation amplitude involves the statistical concepts of variance and standard deviation. The system calculates the deviation of each time interval data point from the location of central tendency; these deviation values ​​reflect the degree to which individual data points deviate from the center. The variance is obtained by summing the squares of all deviations and dividing by the total number of records; the square root of the variance is the standard deviation. The standard deviation represents the degree of dispersion of the data in units of time; a larger value indicates more severe fluctuations in response time and poorer driver response stability.

[0105] In one possible implementation, outlier identification and removal employs a standard deviation-based rule. According to statistical principles, in a normal distribution, approximately 95% of the data falls within the range of the mean plus or minus two standard deviations. The system identifies data points outside this range as outliers, which may be due to driver distraction, system malfunctions, or other special circumstances. After removing outliers, the system recalculates the average of the remaining data; this cleaned average more accurately reflects the driver's typical response time.

[0106] Preferably, the optimized evaluation time is determined directly by using the average response time after removing outliers. This time represents the driver's true response to the shift prompt after eliminating interference factors. Compared to a simple average of the original data, the optimized evaluation time better reflects the actual guidance effect of the system prompts.

[0107] Step S106: Evaluate the improvement of the original response time baseline value by shortening the evaluation time, obtain the effectiveness score of the real prompt guidance response, evaluate the degree of conformity between the proportion of real prompt guidance response behaviors and the predicted behavior standard threshold, and obtain an accurate evaluation output of the motorcycle shift prompt effect.

[0108] The difference between the optimized evaluation time and the original response time baseline is calculated. If the difference is negative and its absolute value exceeds a preset threshold, the improvement ratio is obtained by dividing the absolute value of the difference by the original response time baseline. The improvement ratio is then multiplied by a preset full score to determine the effectiveness score of the real prompt guidance response. Weighting coefficients are set based on the effectiveness score. The proportion of the number of real response sub-records to the total time interval records is calculated to obtain the rate of change of the correction magnitude relative to the initial value. Combined with the strength value of the early gear shifting habit and the proportion of proactive prediction operation records to the total operation records, the various indicators are weighted and summed according to the weighting coefficients to form a comprehensive evaluation report containing various quantitative indicators. The proportion of the number of real response sub-records to all records is extracted from the comprehensive evaluation report. The absolute value of the difference between this proportion and the prediction behavior standard threshold is calculated. If the absolute value of the difference is less than 20% of the threshold, the system evaluation result is deemed to meet the standard. The system recognition accuracy is determined based on the sign of the difference. The system's recognition accuracy is used to adjust the credibility of each indicator in the comprehensive evaluation report. The adjusted indicator values ​​are compared with preset multi-level evaluation thresholds to determine the level of the motorcycle shift prompt effect, thus obtaining an accurate evaluation output of the motorcycle shift prompt effect.

[0109] Optionally, the weighting coefficients are set based on the validity score range. For example, a score above 80 has a weight of 1, a score between 60 and 80 has a weight of 0.8, and a score below 60 has a weight of 0.5. The initial value of the correction magnitude is defined as the average response time of the first recorded data. The calculation of the actual response percentage is performed only once when the comprehensive evaluation report is generated, and subsequent steps directly use the results.

[0110] Specifically, in one implementation, the improvement is calculated based on a comparative analysis of the optimized evaluation time and the original response time baseline. The original response time baseline is a standard response time obtained through statistical analysis of a large amount of normal driving data in the initial stage of the system, typically between 600 and 1000 milliseconds. The optimized evaluation time is the actual response time obtained after outlier removal and statistical processing. When the optimized evaluation time is 400 milliseconds and the original baseline is 800 milliseconds, the difference is 400 minus 800 equals -400 milliseconds, indicating a reduction in response time of 400 milliseconds. The response speed improvement ratio is the difference divided by the original baseline, i.e., 400 divided by 800 equals 0.5, which translates to a 50% percentage improvement, indicating a 50% increase in response speed. This percentage is used as the improvement ratio to reflect the actual degree of improvement in driver response speed provided by the shift prompt system. The calculation of the effectiveness score for the actual prompt guidance response involves the conversion relationship between the improvement ratio and the preset maximum score. The system's preset maximum score is 100 points; when the improvement ratio reaches or exceeds 50%, the effectiveness score is close to the maximum. Optionally, the system presets a maximum score of 100 points. When the improvement rate reaches or exceeds 50%, the effectiveness score is between 90 and 100 points. The score calculation uses a non-linear mapping: when the improvement rate is between 0 and 20%, the score increases slowly; between 20% and 50%, the score increases rapidly; and above 50%, the score approaches the maximum and tends to stabilize. This non-linear mapping method can more accurately reflect the impact of different improvement levels on the actual driving experience, avoiding evaluation biases that may arise from a simple linear relationship. In one embodiment, the scoring standard is set based on the non-linear relationship between the improvement rate and the improvement in driving experience. The 50% corresponding to the 90-100 point range is based on extensive driving data analysis, reflecting a significant improvement in experience. The calculation formula is as follows: when the improvement rate is less than 20%, the score is the rate multiplied by 2; when it is 20% to 40%, the score is 40 plus the rate multiplied by 1.5; when it exceeds 40%, the score is 80 plus the rate multiplied by 0.5, with a maximum of 100 points.

[0111] For example, the weighting coefficients are dynamically adjusted based on the effectiveness score. When the effectiveness score exceeds 80, the system considers the shift prompts to be effective. In this case, the weight of the proportion of actual response sub-records is set to 0.4, the weight of the correction magnitude change rate is set to 0.2, the weight of the early shift habit intensity value is set to 0.25, and the weight of the proportion of proactive prediction operation records is set to 0.15. This weighting method highlights the importance of actual response records while taking into account the influence of other dimensions. When the effectiveness score is low, the system will adjust the weighting accordingly, increasing the weight of habit intensity and prediction proportion to more comprehensively evaluate the driver's operating characteristics.

[0112] In one possible implementation, the construction of the comprehensive evaluation report involves the integration and analysis of multiple quantitative indicators. The proportion of the number of true response sub-records directly reflects the proportion of effective responses identified by the system among all operations; the higher this proportion, the better the system's identification accuracy. The rate of change in correction amplitude, calculated by comparing correction amplitude values ​​at different time periods, reflects the stability of the driver's operating habits. The strength of the early gear shifting habit comes from previous statistical analysis and reflects the driver's predictive ability level. The proportion of proactive predictive operation records reflects the driver's driving proficiency from another perspective. The system weights and sums these four indicators according to set weighting coefficients to obtain a comprehensive score, and organizes the raw data, calculation process, and final score into a structured report format. Alternatively, the four indicators—the proportion of the number of true response sub-records, the rate of change in correction amplitude, the strength of the early gear shifting habit, and the proportion of proactive predictive operation records—can be normalized and then weighted and summed according to set weighting coefficients to obtain a comprehensive score.

[0113] Preferably, the accuracy of the system's identification is determined using a deviation analysis method. The proportion of true response sub-records to all records is extracted from the comprehensive evaluation report. Theoretically, this proportion should be close to the predicted behavior standard threshold. The predicted behavior standard threshold is set between 0.3 and 0.4, representing the expected proportion of a driver's predicted behavior under normal circumstances. This threshold is determined based on multiple driving behavior studies and industry standard statistical analysis, combined with expert evaluation, ensuring the parameters are scientific and reasonable. When the absolute value of the difference between the actual proportion and the standard threshold is less than 20% of the threshold, the system determines that the evaluation result meets expectations. A positive difference indicates that the system may over-identify, misclassifying some non-predicted behaviors as predicted; a negative difference indicates that the system under-identifies, missing some predicted behaviors.

[0114] Specifically, the credibility adjustment mechanism corrects various indicators in the comprehensive evaluation report based on the system's recognition accuracy. When the recognition accuracy is high, the indicators remain unchanged or undergo only minor adjustments; when the recognition accuracy is low, the system will make compensatory adjustments to the corresponding indicators according to the direction and magnitude of the deviation.

[0115] For example, if the system tends to over-identify, the percentage of proactively predicted operation records will be appropriately reduced; if it under-identifies, the percentage will be increased accordingly. This adjustment ensures the reliability of the final evaluation results.

[0116] Understandably, the multi-level evaluation thresholds are based on statistical analysis of a large amount of historical data. The shift indicator effectiveness is divided into four levels: Excellent, Good, Satisfactory, and Needs Improvement, each corresponding to a different range of indicators. An adjusted overall score above 85 is Excellent, 70-85 is Good, 55-70 is Satisfactory, and below 55 is Needs Improvement. These thresholds are based on industry-recognized evaluation standards and years of practical experience from an expert group, ensuring the classification is scientific and feasible. The system compares each adjusted indicator value with these thresholds one by one to determine the final effectiveness level.

[0117] For example, in a certain evaluation, the system calculated an improvement rate of 45%, an effectiveness score of 88, a true response sub-record percentage of 68%, a correction magnitude change rate of 12%, an early gear shift habit strength value of 0.75, and a proactive prediction operation percentage of 35%. The true response percentage is directly taken as a decimal of 0.68, the correction magnitude change rate after inverse normalization is 1−12 / 50=0.76, and the proactive prediction percentage is directly taken as 0.35. The weight of the percentage of true response sub-records is set to 0.4, the weight of the correction magnitude change rate is set to 0.2, the weight of the early gear shift habit strength value is set to 0.25, and the weight of the proactive prediction operation record percentage is set to 0.15. After weighted calculation, the comprehensive score is 0.68*0.4+0.76*0.2+0.75*0.25+0.35*0.15.

[0118] In one embodiment, the system also analyzes the differences in evaluation results under different road conditions. In congested urban traffic, drivers shift gears frequently, and their predictive behavior is more pronounced, often resulting in higher evaluation results. On highways, the number of gear shifts decreases, and predictive behavior is relatively less pronounced, potentially leading to lower evaluation results. By identifying these scenario characteristics, the system provides contextual interpretations of the evaluation results, making the final evaluation output more accurate and practical.

[0119] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for evaluating motorcycle gear shift prompts using intelligent algorithms, characterized in that, include: Acquire time interval records related to the driver's gear shifting operation. The time interval records include the time difference between the clutch pre-action moment and the system prompt moment, as well as its positive and negative distribution characteristics. The predicted behavior recognition time window length and the predicted behavior standard threshold are determined based on the time interval records. Determine the proportion of short response times, analyze the deviation of the proportion of short response times from the predicted behavior standard threshold, and distinguish between driver active prediction operation records and passive response prompt operation records; The discrimination threshold is determined by the action connection interval distribution of the active prediction operation record and the passive response prompt operation record. The discrimination threshold is adjusted based on the individual differences actively predicted by the driver, and the time interval records are filtered to obtain a set of real response sub-records; The response time distribution characteristics of the real response sub-record set are statistically analyzed to determine the optimized evaluation time, and the improvement of the optimized evaluation time on the original response time is evaluated. The evaluation result of the shift prompt effect is then output.

2. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, The acquisition of time interval records related to the driver's gear shifting operations includes: The clutch pedal pressure sensor collects the moment when the driver begins to apply force with his foot as the clutch preparation moment. Record the time when the vehicle's gear shift indicator emits an audible and visual signal. Calculate the time difference between the clutch preparatory action moment and the prompt issuance moment, and generate the time interval record containing positive and negative distribution characteristics.

3. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, After acquiring the time interval records related to the driver's gear shifting operation, based on the continuously collected gear shifting operation time difference sequence in the time interval records, the ratio of the number of negative difference occurrences to the total number of occurrences is calculated. The time difference sequence is segmented using the sliding time window method to determine the length of the prediction behavior recognition time window. The median of the positive difference values ​​is extracted from the time interval records as the original response time reference value.

4. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, The step of determining the threshold for predicting behavior based on the time interval records includes: The standard deviation of the time difference within the negative value interval is calculated by using the ratio of the number of occurrences of negative differences in the time interval records to the total number of occurrences. The data density is determined based on the ratio of the standard deviation to the length of the predicted behavior recognition time window. The data density is compared with a preset threshold to generate the predicted behavior standard threshold.

5. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, Determining the proportion of short response time includes: calculating the proportion of negative data points to the total number of records by using the negative time difference data recorded in the time interval as the proportion of short response time.

6. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, After determining the proportion of short response times, the following steps are included: Calculate the difference between the proportion of short response times and the predicted behavior standard threshold; The direction of deviation of the driver's anticipated behavior is determined based on the sign of the difference; If the difference is positive and exceeds the preset deviation threshold, then continuous negative data points are extracted from the negative value interval recorded in the time interval, and the number of data points within the unit time window is counted as the data density value to obtain the density distribution reflecting the concentration of the predicted behavior. Based on the location of the peak region in the density distribution, the distribution ratio of the short response time in different negative sub-intervals is extracted. The ratio of the number of data points in each sub-interval to the total number of negative data points is calculated as the contribution weight. The timing advance of the predicted behavior is obtained by weighted summation of the distribution ratio of each negative sub-interval using the contribution weight.

7. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, The determination of the discrimination threshold based on the action connection interval distribution of the active prediction operation record and the passive response prompt operation record includes: Calculate the ratio of the time lead to the length of the predicted behavior recognition time window; Using the aforementioned proportional coefficient as a boundary standard, operation records with a value less than the product of the proportional coefficient and a preset benchmark value are selected from the action connection interval sequence. The selected operation records are classified according to the sign of the time interval value. Records with negative time interval values ​​are marked as active prediction operation records, and records with positive time interval values ​​are marked as passive response prompt operation records. By calculating the average action connection interval of the active prediction operation record and the average action connection interval of the passive response prompt operation record, typical characteristic values ​​of the two types of operations are determined. Based on the midpoint of the two means and the standard deviation of the action connection interval distribution, a discrimination threshold is generated to distinguish between the two types of operation records.

8. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, After determining the discrimination threshold, the process includes: The change in capacitance of the shift lever touch sensor is detected within the data segment prior to the time of system prompt issuance, filtered from the time interval record. If the capacitance value exceeds the preset touch threshold, the moment when the driver's hand touches the gear shift lever is recorded, and the time difference between the moment the gear shift lever is touched and the moment the system prompts the system is activated is calculated. If the time difference is negative and its absolute value exceeds a preset threshold, it is determined that the hand pre-touching has been completed. By using the hand pre-touch completion state, the clutch pedal pressure curve and shift lever position signal for the corresponding time period are extracted from the action connection interval data to determine the sequence of clutch disengagement and shift lever positioning. If the preset coordination threshold is met, the driver's active gear shifting operation mode is determined by combining road slope data and real-time vehicle speed values.

9. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, The step of adjusting the discrimination threshold of the proactive prediction operation record based on individual differences in driver proactive prediction, and filtering the time interval records to obtain the set of real response sub-records, includes: Obtain the intensity values ​​of early gear shifting habits of different drivers, and calculate the difference ratio between each driver's habit intensity value and the average habit intensity value of all drivers; The individual difference correction factor is determined based on the distribution range of the difference ratio; Calculate the ratio of the predicted behavior advance amount to the length of the predicted behavior recognition time window; The correction magnitude is determined by multiplying the ratio by the individual difference correction factor, and the upper and lower boundary values ​​of the discrimination threshold of the active prediction operation record are adjusted accordingly. The time interval records are filtered one by one according to the adjusted boundary, and the records located between the lower boundary value and the upper boundary value are retained. Check whether the shift operation corresponding to the filtered record contains a complete process. If the condition is met, include it in the set of real response sub-records.

10. The motorcycle shift prompt evaluation method using intelligent algorithms as described in claim 1, characterized in that, The step of performing response time distribution characteristic statistics on the set of real response sub-records to determine the optimized evaluation duration, and evaluating the improvement of the optimized evaluation duration on the original response time, includes: Sort the time interval data in the set of real response sub-records, and calculate the median and mean of the time interval data; The symmetry of the data distribution is determined by comparing the difference between the median and the mean, and the location of central tendency is analyzed. Based on the central tendency position, calculate the deviation of each time interval data from the central tendency position, and obtain the standard deviation to measure the discrete fluctuation amplitude; The optimized evaluation duration is determined by removing outliers and recalculating the average value based on the central trend position and the discrete fluctuation amplitude; The extent of improvement is assessed based on the difference between the optimized assessment time and the original response time baseline.

Images