A big data analysis automobile service demand dynamic matching scheduling management method
By using big data analysis and real-time video interactive control methods, the problem of time and space mismatch between supply and demand of car services in the traditional scheduling mode has been solved, achieving more accurate resource matching and improved management efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- QINGMIN DIGITAL TECHNOLOGY (QINGDAO) TECHNOLOGY SERVICE CO LTD
- Filing Date
- 2026-03-18
- Publication Date
- 2026-06-16
AI Technical Summary
In traditional dispatching models, the mismatch between the supply of car services and user demand in time and space leads to insufficient matching accuracy, failure to predict in a timely manner, and failure to consider the impact of changes in road conditions and environment on car owners' needs.
The real-time video interactive control method based on big data matches target vehicle preference and demand information with resource detection, divides road segments for road condition and environmental information analysis, statistically analyzes the difference in emotion index and correlation influence characteristic coefficients, predicts changes in driver emotions, and performs dynamic scheduling.
It improves the timeliness and predictability of resource matching, enhances resource utilization and management efficiency, and reduces waiting time for car owners.
Smart Images
Figure CN122222283A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of dynamic demand scheduling management technology, and in particular to a method for dynamic matching and scheduling management of automotive service demand based on big data analysis. Background Technology
[0002] The core contradiction in car services (including ride-hailing, taxis, roadside assistance, maintenance, shared mobility, and logistics) is the mismatch between "service supply" and "user demand" in time and space.
[0003] The inefficiency of traditional scheduling models has become the core driving force behind the intervention of big data analysis. For example, in traditional technology, resource matching is often based on a comprehensive judgment of the driver's resource needs, the distance between them, and the road conditions. However, it does not further judge the impact of environmental factors or the correlation between the passengers and the driver's needs. Furthermore, traditional technology does not make timely predictions for the changes in the diverse situations of different road sections, which can easily lead to insufficient matching accuracy and other pain points. Summary of the Invention
[0004] To overcome the shortcomings of the prior art, this application provides a real-time video interactive control method based on big data.
[0005] This application provides a real-time video interactive control method based on big data, the method comprising: Step S1: Based on the target vehicle's target preference information, detect matching resources that match the target preference information, output initial matching resources, divide the route between the target vehicle and the initial matching resources into road segments, extract the front sub-segment and the back sub-segment, detect the road condition and environmental information of the front sub-segment, and obtain road condition feature one and environmental feature one. Step S2: If there are no other passengers in the target vehicle, output the non-passenger condition. Based on the non-passenger condition, statistically analyze the difference between the driver's emotional index of adjacent sub-segments in the previous test sub-segment to obtain the difference of the test emotional index. Based on the correlation influence of the normal stability of road condition feature 1 and environmental feature 1 with the change of driver's emotional state, statistically analyze the first correlation influence feature coefficient between road condition feature 1, environmental feature 1 and the difference of the test emotional index. Step S3: If there are other passengers in the target vehicle, output the passenger carrying conditions. Based on the passenger carrying conditions, statistically analyze the differences between the emotional indices of other passengers in adjacent sub-segments of the preceding sub-segment to obtain the auxiliary emotional index difference. Based on the correlation between the changes in the emotional state of other passengers, the normal stability of road condition feature one and environmental feature one, and the changes in the emotional state of the vehicle owner, statistically analyze the second correlation influence feature coefficient. Based on the first correlation influence feature coefficient, the second correlation influence feature coefficient, and the road condition feature two and environmental feature two of the following sub-segment, predict the changes in the emotional index of the vehicle owner to obtain the preprocessed emotional prediction index. Then, schedule the initial matching resources and output the demand dynamic matching scheduling result.
[0006] Preferably, the target vehicle's target preference information is obtained, and matching resources that meet the requirements are detected based on the target preference information, and initial matching resources are output. Obtain the current location of the target vehicle, calculate the test route between the location of the nearest resource in the initial matching resources and the current location, and divide the test route into several equal segments to obtain the test sub-segments. A preset analysis segment threshold is set, and the sub-segment to be tested is marked as the pre-test sub-segment and the post-test sub-segment based on the analysis segment threshold; The road condition information and environmental information of each sub-segment in the aforementioned pre-test sub-segment are detected in real time to obtain road condition feature one and environmental feature one.
[0007] Preferably, the emotional and behavioral feature information of the car owner at the current location is collected, and the initial emotional index of the car owner is evaluated by a CNN-LSTM model based on the emotional and behavioral feature information. The driver sentiment index of each sub-segment in the preceding test sub-segment is statistically analyzed to obtain the sentiment index set of the preceding test segment; The difference between the sentiment indices of the previously tested road segments and the sentiment indices of the adjacent sub-road segments is calculated to obtain the difference in the sentiment index of the target road segment. According to the non-passenger-carrying conditions, if the first road condition feature is unconventional and stable, and the first environmental feature is conventional and stable, then the difference value between the road condition feature data of adjacent sub-road segments in the first road condition feature is calculated to obtain the road condition change feature value to be measured. Based on the correlation influence characteristics between road condition feature 1 and driver's emotional state, the first correlation influence coefficient 1 between the difference between the road condition feature value to be tested and the emotional index to be tested is calculated.
[0008] Preferably, if road condition feature one is normal and stable, and environmental feature one is non-normal and stable, then the difference value between the environmental feature data of adjacent sub-road segments in environmental feature one is calculated to obtain the environmental change feature value to be measured. Based on the correlation and change characteristics between environmental feature one and the emotional state of the car owner, the first correlation coefficient two is calculated between the difference between the measured environmental change feature value and the measured emotional index. If road condition feature 1 is unconventional and stable, and environmental feature 1 is also unconventional and stable, then based on the correlation influence of road condition feature 1 and environmental feature 1 on the change characteristics of the driver's emotional state, the first correlation influence coefficient 3 is calculated for the difference between the test road condition change feature value, the test environmental change feature value and the test emotional index. The first correlation influence coefficient one, the first correlation influence coefficient two, and the first correlation influence coefficient three are combined to form the first correlation influence characteristic coefficient.
[0009] Preferably, based on the passenger carrying conditions, the second set of emotional and behavioral characteristic information of other passengers at their current location is collected, and based on the difference between the second set of emotional and behavioral characteristic information and the test emotional index, the difference of the auxiliary test emotional index of other passengers is calculated. If the first road condition feature is unconventional and stable, and the first environmental feature is conventional and stable, then based on the correlation influence of the first road condition feature and the emotional state of other passengers on the emotional state of the driver, the second correlation influence coefficient one between the test road condition change feature value, the auxiliary emotional index difference, and the test emotional index difference is calculated.
[0010] Preferably, if road condition feature one is normal and stable, and environmental feature one is non-normal and stable, then based on the correlation influence of environmental feature one and the emotional state of other passengers on the emotional state of the car owner, the second correlation influence coefficient two between the measured environmental feature value, the difference of auxiliary emotional index and the difference of the measured emotional index is calculated. If road condition feature 1 is unconventional and stable, and environmental feature 1 is also unconventional and stable, then based on the correlation influence of road condition feature 1, environmental feature 1, and the emotional state of other passengers on the emotional state of the car owner, the following statistical values are calculated: the test road condition change feature value, the test environmental change feature value, the auxiliary test emotional index difference, and the second correlation influence coefficient of the test emotional index difference. The second correlation influence coefficient one, the second correlation influence coefficient two, and the second correlation influence coefficient three are combined to form the second correlation influence characteristic coefficient; The first correlation influence characteristic coefficient and the second correlation influence characteristic coefficient are combined to form the preprocessing influence coefficient.
[0011] Preferably, the road condition information and environmental information of each sub-segment in the post-test sub-segment are detected in real time to obtain road condition feature two and environmental feature two; Based on the second road condition feature, the second environmental feature, and the preprocessing influence coefficient, the change in the car owner's emotional index is predicted to obtain the preprocessed emotional prediction index. A preset emotion index judgment threshold is set. If the preprocessed emotion prediction index is greater than or equal to the emotion index judgment threshold, the information on changes in car owner demand is output. Based on the location characteristics corresponding to the changes in the vehicle owner's demand, the initial matching resources are scheduled to match the target vehicle to the resource location with the shortest travel time, and the dynamic matching and scheduling results of the demand are output.
[0012] Compared with the prior art, the present invention has the following characteristics and beneficial effects: By initially matching resources based on the target vehicle's desired needs, an initial matching resource is obtained. However, this approach faces the significant possibility that changes in road conditions and environment during the target vehicle's journey could increase the waiting time for resources, causing delays. Traditional methods often directly match resources based on the overall road conditions and environmental characteristics without further detailed analysis of sub-segments. The presence or absence of passengers in the target vehicle also influences the driver's needs. Therefore, a differentiated analysis is conducted for the two scenarios: one where the target vehicle is carrying passengers, and another where it is not. The analysis considers the stability of road conditions and environmental characteristics when passengers are present, examining their impact on the driver's emotional state. If the road conditions and environmental characteristics under passenger conditions are generally stable, an analysis of the impact on the driver's emotional index is conducted. Furthermore, the analysis also considers the emotional index changes of other passengers after being affected by changes in road conditions and environmental characteristics, thus shifting the focus to analyzing the impact on the driver's emotional index. This allows for more accurate and effective demand-based scheduling based on changes in the driver's emotional index. To improve the timeliness of dynamic resource matching and scheduling, the test route is divided into several equal sub-segments, and pre-test and post-test sub-segments are extracted. The pre-test sub-segments are used for feature statistics and analysis to provide reliable, regular reference feature information for the post-test sub-segments, i.e., for prediction. This improves the timeliness and predictability of dynamic demand matching and scheduling, enhancing resource utilization and the efficiency of resource matching management. Attached Figure Description
[0013] Figure 1 This embodiment mainly illustrates a step-by-step flowchart of a real-time video interactive control method based on big data. Detailed Implementation
[0014] The present invention will be further described in detail below with reference to the following embodiments.
[0015] Reference Figure 1 A real-time video interactive control method based on big data, the method includes the following steps: Step S1: Based on the target vehicle's target preference information, detect matching resources that match the target preference information, output initial matching resources, divide the route between the target vehicle and the initial matching resources into road segments, extract the front test sub-segment and the back test sub-segment, detect the road condition and environmental information of the front test sub-segment, and obtain road condition feature one and environmental feature one.
[0016] Step S2: If there are no other passengers in the target vehicle, output the non-passenger condition. Based on the non-passenger condition, statistically analyze the differences between the driver's emotional index of adjacent sub-segments in the previous test sub-segment to obtain the difference of the test emotional index. Based on the correlation between the normal stability of road condition feature 1 and environmental feature 1 and the changes in driver's emotional state, statistically analyze the first correlation influence characteristic coefficient between road condition feature 1, environmental feature 1 and the difference of the test emotional index.
[0017] Step S3: If there are other passengers in the target vehicle, output the passenger carrying conditions. Based on the passenger carrying conditions, statistically analyze the differences between the emotional indices of other passengers in adjacent sub-segments of the preceding sub-segment to obtain the auxiliary emotional index difference. Based on the correlation between the changes in the emotional state of other passengers, the normal stability of road condition feature one and environmental feature one, and the changes in the emotional state of the vehicle owner, statistically analyze the second correlation influence feature coefficient. Based on the first correlation influence feature coefficient, the second correlation influence feature coefficient, and the road condition feature two and environmental feature two of the following sub-segment, predict the changes in the emotional index of the vehicle owner to obtain the preprocessed emotional prediction index. Then, schedule the initial matching resources and output the dynamic matching scheduling result of the demand.
[0018] Specifically, for example, by initially matching resources based on the target vehicle's desired needs, an initial matching resource is obtained. However, there is a high probability that the target vehicle will experience changes in road conditions and environment during its journey, leading to increased waiting time for resources and causing delays. Furthermore, traditional techniques often directly match resources based on the overall road conditions and environmental characteristics of the route without further detailed analysis of sub-segments. The presence or absence of passengers in the target vehicle also has a certain correlation with changes in the driver's needs. A differentiated analysis is needed for the two scenarios: one is whether the road conditions and environmental characteristics are generally stable when passengers are present, which then influences the driver's emotional index; the other is... If the road conditions and environmental characteristics are stable under passenger loads, an analysis of the impact on the driver's emotional index is conducted. This analysis also includes the impact of changes in the emotional index of other passengers on road conditions and environmental characteristics, further shifting the focus to the driver's emotional index. This allows for more accurate and effective demand-based scheduling based on changes in the driver's emotional index. To improve the timeliness of dynamic resource matching, the test route is divided into several equal sub-segments, and pre-test and post-test sub-segments are extracted. The pre-test sub-segments are used for feature statistics and analysis to provide reliable, regular reference feature information for the post-test sub-segments, i.e., for prediction. This improves the timeliness and predictability of dynamic demand matching, enhancing resource utilization and the efficiency of resource matching management.
[0019] The specific step S1 includes the following sub-steps: Obtain the target vehicle's target preference information, detect matching resources that meet the requirements based on the target preference information, and output the initial matching resources.
[0020] Obtain the current location of the target vehicle, calculate the test route between the location of the nearest resource in the initial matching resources and the current location, and divide the test route into several equal segments to obtain the test sub-segments.
[0021] Preset the analysis segment threshold, and mark the sub-segment to be tested as the pre-test sub-segment and the post-test sub-segment based on the analysis segment threshold.
[0022] Real-time detection of road condition and environmental information of each sub-segment in the preceding sub-segment is performed to obtain road condition feature one and environmental feature one.
[0023] Specifically, initial resource matching (data collection authorization and startup: when the car owner registers the APP or WeChat mini program, location information and sensor usage authorization are obtained; the APP integrates the Gaode / Baidu Map SDK, enables GPS positioning (default accuracy within 10 meters), and collects location data every 3 seconds; the phone's accelerometer collects acceleration values every 0.5 seconds, and judges the target vehicle's driving status (stationary / driving / idling) through a threshold (horizontal acceleration > 0.3g and lasting for 2 seconds); data synchronization and completion: the WeChat mini program synchronizes the car owner's reservation information and service preferences in real time through service call interfaces; the APP adopts a "real-time + batch" hybrid transmission strategy: driving In a stationary state, location and acceleration data are uploaded every 5 seconds; in a stationary state, data is uploaded in batches every 30 seconds. In case of network failure, data is cached locally (maximum of 100 records), and re-uploaded in timestamp order after network recovery. Real-time traffic data is obtained through the map SDK and fused with target vehicle location data to generate a dynamic driving trajectory. Interface integration and adaptation: The store ERP system is integrated via a RESTful API, configured with a scheduled synchronization task (retrieving workstation status and technician shift data every minute); the POS system has real-time data push functionality enabled, pushing order amount and payment status via WebHook within 1 second of transaction completion; the appointment management system uses MySQL. Direct database connection: A read-only account is established to retrieve appointment order data (including service type, appointment time, and vehicle owner information). Asynchronous data synchronization: A RabbitMQ message queue is introduced to encapsulate data from various business systems into standardized messages (including data type, timestamp, and checksum), which are then routed to different queues according to categories such as "order data," "resource status data," and "transaction data." The consumer side uses an ACK confirmation mechanism to perform integrity verification (checksum comparison) upon receiving data. If the verification passes, the data is stored in the database; otherwise, it is re-queued (up to 3 retries) to ensure no data loss or duplication. This includes preference demand prediction (data preprocessing: collecting historical order data from the past year, constructing...). Construct a "user-store" rating matrix (rating = loyalty rating × 0.4 + service satisfaction rating × 0.6), and fill in the missing values in the matrix (using a weighted average of user mean + store mean)); Matrix decomposition and training: Use the SVD++ algorithm to decompose the rating matrix, set the latent vector dimension to 20, the learning rate to 0.001, and the number of iterations to 100, minimizing the mean squared error between the predicted rating and the actual rating (target MSE < 0.5); Preference prediction: Input the current car owner's demand characteristics and store resource characteristics, and use the trained model to predict the car owner's preference score (0-10 points) for each store, and filter stores with preference scores ≥ 6 points into the candidate set.Resource matching: Real-time route time calculation: Call the Gaode / Baidu Map SDK, input the driver's current location and the store location, obtain 3 candidate routes, and calculate the real-time travel time of each route using the Dijkstra algorithm (integrating traffic flow data, the time coefficient for congested road sections is increased by 1.2-2.0 times); Workstation time window constraint verification: Obtain the reserved time slots and current occupancy status of each workstation in the store, generate available time windows for workstations (such as 9:00-10:30, 11:00-12:00), and verify whether the driver's expected arrival time overlaps with the available time window. If the overlap time is greater than or equal to the estimated service time, the store is retained; Matching and filtering: Sort by the sum of "travel time + waiting time" in ascending order, and filter the top 20% of stores to enter the next round of matching. This involves comprehensively obtaining initial matching resources), the route to be tested (referring to the required navigation route), sub-segments to be tested (which can be divided into several sub-segments of equal average length based on the average length of multiple congested road segments), preset analysis segment thresholds (if it is 1 / 3, it can be set manually), pre-test sub-segments and post-test sub-segments (pre-test sub-segments refer to sub-segments whose length accounts for 1 / 3 of the total length of the route to be tested, and so on, post-test sub-segments refer to the remaining sub-segments whose length accounts for 2 / 3 of the total length of the route to be tested), road condition feature one and environmental feature one (detection of road condition features: such as using high-definition network cameras installed along the road, the cameras collect dynamic video of the road area in real time, and the backend uses computer vision algorithms to detect, track, and count vehicles in the video, while calculating parameters such as vehicle spacing, queue length, and driving speed; based on the algorithm to identify congestion features, such as vehicle queue length exceeding the threshold or average speed lower than the set value, directly determining the congestion level; and through the number of vehicles in the area and The ratio of road area quantifies vehicle density. If multi-dimensional parameters (queue length, vehicle speed, vehicle density, traffic flow) are detected through vehicle detection or tracking, each basic indicator is first standardized (eliminating dimensional differences). Then, a weighted sum is calculated based on the indicator's contribution to congestion (weight), yielding a comprehensive congestion index. Indicators with different dimensions (e.g., density is "vehicles / m²", speed is "km / h") are converted to dimensionless values of 0-1. If an indicator is negative (e.g., higher density means more congestion, larger values mean more congestion) – such as the X standard. = (X-Xmin) / (Xmax-Xmin), if the indicator is a positive indicator (e.g., the higher the speed, the smoother the traffic, the larger the value, the smoother the traffic) -- such as X standard = (Xmax-X) / (Xmax-Xmin), where Xmin and Xmax are the historical extreme values of the indicator (e.g., the maximum / minimum density of the road, the highest / lowest speed), weight determination: assign weights to each indicator to reflect its contribution to congestion. The weights can be calculated using the analytic hierarchy process (AHP). Multiply the standardized indicators by their weights and sum them to obtain the congestion index.Environmental characteristics: Air temperature, humidity, wind speed, wind direction, rainfall (rain gauge), visibility (fog, haze), light intensity, and atmospheric pressure can be detected through road weather monitoring stations. Subsequent calculations can be made based on the statistical steps of the road congestion index.
[0024] The specific step S2 includes the following sub-steps: Collect the emotional and behavioral features of the car owner at their current location. Based on these features, use a CNN-LSTM model to evaluate the car owner's initial emotional index.
[0025] The driver sentiment index of each sub-segment in the pre-test sub-segment is statistically analyzed to obtain the sentiment index set of the pre-test sub-segment.
[0026] The difference between the sentiment indices of adjacent sub-segments of the previously tested road segment is calculated to obtain the difference in the sentiment index of the target road segment.
[0027] Based on the non-passenger-carrying conditions, if road condition feature one is unconventional and stable, and environmental feature one is conventional and stable, then the difference value between the road condition feature data of adjacent sub-road segments in road condition feature one is calculated to obtain the road condition change feature value to be measured.
[0028] Based on the correlation between road condition feature 1 and driver's emotional state, the first correlation coefficient 1 between the difference between the road condition feature value to be tested and the emotional index to be tested is calculated.
[0029] If road condition feature one is normal and stable, and environmental feature one is non-normal and stable, then the difference value between the environmental feature data of adjacent sub-road segments in environmental feature one is calculated to obtain the environmental change feature value to be measured.
[0030] Based on the correlation and change characteristics between environmental feature one and the emotional state of the car owner, the first correlation coefficient two is calculated between the difference between the environmental feature value to be measured and the emotional index to be measured.
[0031] If road condition feature 1 is unconventional and stable, and environmental feature 1 is also unconventional and stable, then based on the correlation influence of road condition feature 1 and environmental feature 1 on the change characteristics of the driver's emotional state, the first correlation influence coefficient 3 is calculated for the difference between the test road condition change feature value, the test environmental change feature value and the test emotional index.
[0032] The first correlation influence coefficient 1, the first correlation influence coefficient 2, and the first correlation influence coefficient 3 are combined to form the first correlation influence characteristic coefficient.
[0033] Specifically, information such as emotional and behavioral characteristics includes facial expressions (such as frowning, smiling, and changes in eye contact, detected by an in-vehicle integrated camera) and micro-head movements (such as frequent head turning and head lowering) which are the core outward manifestations of emotional states, detected by an in-vehicle integrated camera), body movements (such as changes in grip strength and body tension) and driving behavior (such as accelerator / brake force and steering wheel operation) which are indirect manifestations of emotions, detected by a steering wheel sensor and an in-vehicle integrated camera), physiological signals (wearable physiological monitoring devices (wristbands / smartwatches) that collect heart rate, heart rate variability, blood oxygen saturation, and skin conductance), and voice characteristics (tone, speed, and pitch of speech). Features such as quantity and pauses are important carriers of emotion, such as those collected through microphone arrays (standard in-vehicle equipment); for example, using CNN-LSTM multimodal fusion: CNN processes visual features (face, action), LSTM processes temporal features (physiological signals, speech, driving behavior), and fusion is performed through feature splicing or attention mechanisms to output an emotion index, i.e., the initial emotion index), the difference between the emotion indices to be tested (if the previous test sub-segment contains L1, L2, Ln, and the detected emotion indices of the driver in each sub-segment are Q1, Q2, Qn, then Q2-Q1, Q3-Q2, and so on), if the road condition feature is not normal (for example, the index value is used to classify congestion levels (level classification: based on the index value)). Congestion levels are categorized (e.g., 0-0.2 smooth, 0.2-0.5 basically smooth, 0.5-0.8 lightly congested, 0.8-1 heavily congested; values less than or equal to 0.5 are considered normal and stable). Environmental characteristic one is considered unconventional and stable (and so on). The road condition change characteristic value to be measured (i.e., the difference between the road condition congestion index and the subsequent environmental change characteristic values to be measured) is used. The first correlation influence coefficient one (e.g., the road condition change characteristic value to be measured is used as the value on the x-axis, and the difference between the emotional index to be measured is used as the value on the y-axis; the correlation influence change characteristic curve between the road condition change characteristic value to be measured and the difference between the emotional index to be measured is calculated, and the slope (average) is the first correlation influence coefficient one). Coefficient 2 (based on the first correlation influence coefficient 1, and so on), First correlation influence coefficient 3 (based on the first correlation influence coefficient 1, and so on; it should be noted that: if we establish y=ax+bz, where y refers to the difference in the tested sentiment index, a refers to the tested road condition change characteristic value, x refers to the influence factor between the tested road condition change characteristic value and the difference in the tested sentiment index, b refers to the tested environmental change characteristic value, and z refers to the influence factor between the tested road condition change characteristic value and the difference in the tested sentiment index, substitute the tested road condition change characteristic value, the tested environmental change characteristic value, and the tested sentiment index difference of each sub-segment in the known pre-test sub-segment into it, and obtain x and z, that is, the first correlation influence coefficient 3).
[0034] The specific step S3 includes the following sub-steps: Based on passenger conditions, collect second information on the emotional and behavioral characteristics of other passengers at their current location. Based on the difference between the second information on emotional and behavioral characteristics and the test emotion index, calculate the difference in auxiliary test emotion index for other passengers.
[0035] If road condition feature 1 is unconventional and stable, and environmental feature 1 is conventional and stable, then based on the correlation influence of road condition feature 1 and the emotional state of other passengers on the emotional state of the driver, the second correlation influence coefficient 1 between the test road condition change feature value, the auxiliary emotional index difference, and the test emotional index difference is calculated.
[0036] If road condition feature 1 is normal and stable, and environmental feature 1 is abnormal and stable, then based on the correlation influence of environmental feature 1 and the emotional state of other passengers on the emotional state of the driver, the second correlation influence coefficient 2 between the measured environmental feature value, the difference in the auxiliary emotional index, and the difference in the measured emotional index is calculated.
[0037] If road condition feature 1 is unconventional and stable, and environmental feature 1 is also unconventional and stable, then based on the correlation influence of road condition feature 1, environmental feature 1, and the emotional states of other passengers on the emotional state of the car owner, the second correlation influence coefficient 3 of the road condition change feature value to be tested, the environmental change feature value to be tested, the difference of the auxiliary emotional index, and the difference of the emotional index to be tested is statistically analyzed.
[0038] The second correlation influence coefficient one, the second correlation influence coefficient two, and the second correlation influence coefficient three are combined to form the second correlation influence characteristic coefficient.
[0039] The first and second correlation influence characteristic coefficients are combined to form the preprocessing influence coefficient.
[0040] Real-time detection of road condition and environmental information of each sub-segment in the post-test sub-segment is performed to obtain road condition feature 2 and environmental feature 2.
[0041] Based on road condition feature 2, environmental feature 2, and preprocessing influence coefficient, the changes in the driver's emotional index are predicted, resulting in a preprocessing emotional prediction index.
[0042] A preset emotion index judgment threshold is set. If the preprocessed emotion prediction index is greater than or equal to the emotion index judgment threshold, the information on changes in the car owner's needs will be output.
[0043] Based on the location characteristics corresponding to the changes in the vehicle owner's demand, the initial matching resources are scheduled to match the target vehicle to the resource location with the shortest travel time, and the dynamic matching and scheduling results of the demand are output.
[0044] Specifically, the following parameters are considered: Auxiliary emotional index difference (based on the difference of the emotional index to be tested, and so on); Second Correlation Influence Coefficient 1; Second Correlation Influence Coefficient 2 (based on the First Correlation Influence Coefficient 3, and so on); Second Correlation Influence Coefficient 3 (e.g., y=ax+bz+cr, where c refers to the difference of the auxiliary emotional index, r refers to the correlation influence factor of the difference of the emotional index to be tested on the difference of the auxiliary emotional index; substituting known data information, a, b, and r are obtained, thus obtaining the Second Correlation Influence Coefficient 3); Preprocessed emotional prediction index (i.e., multiplying the known road condition feature 2, environmental feature 2, and preprocessed influence coefficients accordingly to obtain the preprocessed emotional prediction index); Preset emotional index judgment threshold (set by oneself based on the statistics of historical experience data); if the preprocessed emotional prediction index is greater than or equal to the emotional index judgment threshold, it is determined that the car owner needs to change their demand (i.e., demand not dependent on preference); Demand dynamic matching and scheduling results (based on the location characteristics corresponding to the change in the car owner's demand information; obtaining the target vehicle's GPS trajectory point sequence (sorted by time), marking the starting point S and the ending point E; calculating the vertical distance from all intermediate points in the sequence to line segment SE; finding...). The system extracts the point P with the largest distance. If the distance is greater than a threshold (preset to 10 meters), P is retained, and the trajectory is divided into two segments, SP and PE. If the distance of all intermediate points is less than or equal to the threshold, the intermediate points are deleted, and only S and E are retained. The final compressed trajectory point sequence is obtained, reducing the data volume by more than 60% while maintaining a path accuracy error of ≤15 meters. The system schedules the initially matched resources, matching the target vehicle to the resource location with the shortest travel time, and selects the serviceable resource stores (i.e., the dynamic demand matching and scheduling result). Based on the allocation and movement of resource orders over the past three months, the system collects data from various sources over the past three months. Hourly order volume data for each store is used to remove abnormal data such as holidays and promotions. The data is then categorized by weekday / weekend. An ARIMA model (p=2, d=1, q=2) is used to train an order volume prediction model for each store, minimizing the prediction error (target MAE < 5). The model is run at 23:00 every day to predict the hourly order volume demand for each store the following day. Based on the prediction results, technician schedules are adjusted in advance (e.g., adding 2 technicians during peak order periods) and workstations are reserved (e.g., reserving 4 workstations if 8 orders are predicted between 10:00 and 11:00) to avoid excessive temporary load.
[0045] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.
Claims
1. A dynamic matching and scheduling management method for vehicle service demand of big data analysis, characterized in that, Includes the following steps: Step S1: Based on the target vehicle's target preference information, detect matching resources that match the target preference information, output initial matching resources, divide the route between the target vehicle and the initial matching resources into road segments, extract the front sub-segment and the back sub-segment, detect the road condition and environmental information of the front sub-segment, and obtain road condition feature one and environmental feature one. Step S2: If there are no other passengers in the target vehicle, output the non-passenger condition. Based on the non-passenger condition, statistically analyze the difference between the driver's emotional index of adjacent sub-segments in the previous test sub-segment to obtain the difference of the test emotional index. Based on the correlation influence of the normal stability of road condition feature 1 and environmental feature 1 with the change of driver's emotional state, statistically analyze the first correlation influence feature coefficient between road condition feature 1, environmental feature 1 and the difference of the test emotional index. Step S3: If there are other passengers in the target vehicle, output the passenger carrying conditions. Based on the passenger carrying conditions, statistically analyze the differences between the emotional indices of other passengers in adjacent sub-segments of the preceding sub-segment to obtain the auxiliary emotional index difference. Based on the correlation between the changes in the emotional state of other passengers, the normal stability of road condition feature one and environmental feature one, and the changes in the emotional state of the vehicle owner, statistically analyze the second correlation influence feature coefficient. Based on the first correlation influence feature coefficient, the second correlation influence feature coefficient, and the road condition feature two and environmental feature two of the following sub-segment, predict the changes in the emotional index of the vehicle owner to obtain the preprocessed emotional prediction index. Then, schedule the initial matching resources and output the demand dynamic matching scheduling result. 2.The real-time video interactive control method based on big data according to claim 1, wherein, Step S1 includes: Obtain the target vehicle's target preference information, detect matching resources that meet the requirements based on the target preference information, and output the initial matching resources; Obtain the current location of the target vehicle, calculate the test route between the location of the nearest resource in the initial matching resources and the current location, and divide the test route into several equal segments to obtain the test sub-segments. A preset analysis segment threshold is set, and the sub-segment to be tested is marked as the pre-test sub-segment and the post-test sub-segment based on the analysis segment threshold; The road condition information and environmental information of each sub-segment in the aforementioned pre-test sub-segment are detected in real time to obtain road condition feature one and environmental feature one.
3. The real-time video interactive control method based on big data according to claim 2, characterized in that, Step S2 includes: Collect the emotional and behavioral feature information of the car owner at the current location. Based on the emotional and behavioral feature information, evaluate the car owner's initial emotional index using a CNN-LSTM model. The driver sentiment index of each sub-segment in the preceding test sub-segment is statistically analyzed to obtain the sentiment index set of the preceding test segment; The difference between the sentiment indices of the previously tested road segments and the sentiment indices of the adjacent sub-road segments is calculated to obtain the difference in the sentiment index of the target road segment. According to the non-passenger-carrying conditions, if the first road condition feature is unconventional and stable, and the first environmental feature is conventional and stable, then the difference value between the road condition feature data of adjacent sub-road segments in the first road condition feature is calculated to obtain the road condition change feature value to be measured. Based on the correlation influence characteristics between road condition feature 1 and driver's emotional state, the first correlation influence coefficient 1 between the difference between the road condition feature value to be tested and the emotional index to be tested is calculated.
4. The real-time video interactive control method based on big data according to claim 3, characterized in that, Step S2 also includes: If road condition feature one is normal and stable, and environmental feature one is non-normal and stable, then the difference value between the environmental feature data of adjacent sub-road segments in environmental feature one is calculated to obtain the environmental change feature value to be measured. Based on the correlation and change characteristics between environmental feature one and the emotional state of the car owner, the first correlation coefficient two is calculated between the difference between the measured environmental change feature value and the measured emotional index. If road condition feature 1 is unconventional and stable, and environmental feature 1 is also unconventional and stable, then based on the correlation influence of road condition feature 1 and environmental feature 1 on the change characteristics of the driver's emotional state, the first correlation influence coefficient 3 is calculated for the difference between the test road condition change feature value, the test environmental change feature value and the test emotional index. The first correlation influence coefficient one, the first correlation influence coefficient two, and the first correlation influence coefficient three are combined to form the first correlation influence characteristic coefficient.
5. The real-time video interactive control method based on big data according to claim 4, characterized in that, Step S3 includes: Based on the passenger carrying conditions, collect the second set of emotional and behavioral characteristic information of other passengers at their current location. Based on the second set of emotional and behavioral characteristic information and the difference between the tested emotional index, calculate the difference between the auxiliary emotional index of other passengers. If the first road condition feature is unconventional and stable, and the first environmental feature is conventional and stable, then based on the correlation influence of the first road condition feature and the emotional state of other passengers on the emotional state of the driver, the second correlation influence coefficient one between the test road condition change feature value, the auxiliary emotional index difference, and the test emotional index difference is calculated.
6. The real-time video interactive control method based on big data according to claim 5, characterized in that, Step S3 also includes: If road condition feature 1 is normal and stable, and environmental feature 1 is non-normal and stable, then based on the correlation influence of environmental feature 1 and the emotional state of other passengers on the emotional state of the car owner, the second correlation influence coefficient 2 between the measured environmental feature value, the difference of auxiliary emotional index and the difference of the measured emotional index is calculated. If road condition feature 1 is unconventional and stable, and environmental feature 1 is also unconventional and stable, then based on the correlation influence of road condition feature 1, environmental feature 1, and the emotional state of other passengers on the emotional state of the car owner, the following statistical values are calculated: the test road condition change feature value, the test environmental change feature value, the auxiliary test emotional index difference, and the second correlation influence coefficient of the test emotional index difference. The second correlation influence coefficient one, the second correlation influence coefficient two, and the second correlation influence coefficient three are combined to form the second correlation influence characteristic coefficient; The first correlation influence characteristic coefficient and the second correlation influence characteristic coefficient are combined to form the preprocessing influence coefficient.
7. The real-time video interactive control method based on big data according to claim 6, characterized in that, Step S3 also includes: The road condition information and environmental information of each sub-segment in the post-test sub-segment are detected in real time to obtain road condition feature two and environmental feature two. Based on the second road condition feature, the second environmental feature, and the preprocessing influence coefficient, the change in the car owner's emotional index is predicted to obtain the preprocessed emotional prediction index. A preset emotion index judgment threshold is set. If the preprocessed emotion prediction index is greater than or equal to the emotion index judgment threshold, the information on changes in car owner demand is output. Based on the location characteristics corresponding to the changes in the vehicle owner's demand, the initial matching resources are scheduled to match the target vehicle to the resource location with the shortest travel time, and the dynamic matching and scheduling results of the demand are output.