Intelligent guiding method, system and device and storage medium

Abstract

The invention provides an intelligent guiding method, system and device and a storage medium, and the method comprises the steps: obtaining user historical behavior data from a navigation system database, and generating a user preference feature vector; on the basis of a real-time monitoring mechanism, a real-time traffic state parameter matrix is obtained, weight configuration in path planning is dynamically adjusted according to the real-time traffic state parameter matrix and the user preference feature vector, and an optimal weight configuration scheme is generated; road network topology structure data are obtained, a candidate path set is generated, the candidate path set is comprehensively scored according to the optimal weight configuration scheme, and a path priority ranking list is generated; and executing dynamic path adjustment according to the path switching opportunity judgment condition, and generating a final optimized navigation path scheme. According to the method, the real-time traffic state parameters and the user preference characteristics are combined, the path weight configuration is dynamically adjusted, and it is ensured that the recommended path not only conforms to the habits of the user, but also can always adapt to the current road condition.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular, to an intelligent guidance method, system, device, and storage medium. Background Art

[0002] When traditional navigation systems plan routes, they often adopt fixed algorithms and evaluation criteria, making it difficult to adapt to the personalized needs of different users and the dynamically changing traffic environment. This results in a large difference between the recommended route and the actual preferences of users, reducing the user experience. At the same time, due to the lack of in-depth analysis and learning of users' historical behaviors, the system cannot accurately grasp users' travel habits and preference characteristics, making it difficult to provide truly personalized navigation services. In addition, in the complex and ever-changing urban traffic environment, the acquisition and processing of real-time traffic information are lagging, affecting the timeliness and accuracy of navigation decisions. When encountering sudden situations such as traffic congestion and accidents, the system is difficult to quickly make adjustments and optimizations, easily causing users to get into a worse traffic condition. On the other hand, when existing systems evaluate and recommend routes, they often only consider a single factor such as distance or time, ignoring multi-dimensional indicators such as road comfort and landscape beauty that users may care about, and it is difficult to meet the comprehensive needs of users. These problems comprehensively lead to low accuracy and user satisfaction of the navigation system, and it is difficult to provide users with a truly intelligent and personalized travel experience. Summary of the Invention

[0003] To solve the above problems, in a first aspect, the present invention provides an intelligent guidance method, mainly including: Obtain user historical behavior data from the navigation system database, and generate a user preference feature vector based on the user historical behavior data; set a personalized traffic flow threshold according to the user preference feature vector, start a real-time monitoring mechanism, collect current traffic state data, and generate a real-time traffic state parameter matrix; for the real-time traffic state parameter matrix and the user preference feature vector, use the gradient descent algorithm to dynamically adjust the time cost weight, distance cost weight, and congestion cost weight in route planning, and generate an optimal weight configuration scheme; obtain road network topology structure data, generate a candidate route set, comprehensively score the candidate route set according to the optimal weight configuration, and generate a route priority ranking list; monitor the real-time traffic state of the recommended route according to the route priority ranking list, if an anomaly is detected, generate an alternative route set and calculate the route switching timing judgment condition, and perform dynamic route adjustment according to the route switching timing judgment condition to generate a final optimized navigation route plan.

[0004] Further, the obtaining user historical behavior data from the navigation system database and generating a user preference feature vector includes: Extract user historical behavior data from the navigation system database, where the user historical behavior data includes travel time distribution and route selection records; Perform dimensionality reduction processing on the user historical behavior data, extract the main feature components, and determine the key patterns of user behavior; Group user behaviors according to the key patterns, and divide them into time preferences, distance preferences, and congestion preferences; Construct a user preference feature vector based on the time preference component, the distance preference component, and the congestion preference.

[0005] Furthermore, set a personalized traffic flow threshold according to the user preference feature vector, start a real-time monitoring mechanism, collect current traffic state data, and generate a real-time traffic state parameter matrix, including: Classify users through the user preference feature vector to determine the personalized traffic flow threshold; If the personalized traffic flow threshold is determined, start the real-time monitoring mechanism, collect current traffic flow, road congestion index, and weather environment data, and construct a real-time acquisition data set; Extract the traffic flow change trend, congestion index fluctuation characteristics, and weather impact weight according to the real-time acquisition data set; If the traffic flow component exceeds the personalized traffic flow threshold, trigger the dynamic analysis module to generate a real-time traffic state parameter matrix including a passing speed component, a congestion index component, and a weather impact component.

[0006] Furthermore, for the real-time traffic state parameter matrix and the user preference feature vector, use the gradient descent algorithm to dynamically adjust the time cost weight, distance cost weight, and congestion cost weight in the path planning to generate an optimal weight configuration scheme, including: Based on the real-time traffic state parameter matrix, obtain data related to the time cost, distance cost, and congestion cost of the current road to obtain the initial cost distribution characteristics; Determine a comprehensive cost calculation model through the initial cost distribution characteristics combined with the user preference feature vector; Use the gradient descent algorithm to dynamically adjust the adjustment amplitudes of the time cost weight, distance cost weight, and congestion cost weight to generate a preliminary weight configuration combination; If the comprehensive cost of the preliminary weight configuration combination exceeds the preset threshold, update the preliminary weight configuration combination to generate an optimal weight configuration scheme.

[0007] Furthermore, obtain the road network topology structure data, generate a candidate path set, and comprehensively score the candidate path set according to the optimal weight configuration to generate a path priority ranking list, including: Extract the road network topology data and node connection relationship data from the preset database to construct an initial path network model; According to the initial path network model, use the graph traversal algorithm to generate a set of candidate paths; Comprehensively score the set of candidate paths according to the optimal weight configuration to generate a path priority sorted list.

[0008] Further, monitoring the real-time traffic status of the recommended path according to the path priority sorted list, if an abnormality is detected, generating an alternative path set and calculating the path switching timing judgment condition, and performing dynamic path adjustment according to the path switching timing judgment condition to generate a final optimized navigation path plan, including: Obtain the real-time traffic status data, extract the real-time traffic status parameter matrix, and analyze the road condition change trend; If the road condition change trend shows abnormal fluctuations, it is determined that there is a sudden traffic event; if the congestion probability value exceeds the preset threshold, a sudden traffic event identifier is generated; Extract the alternative path set according to the sudden traffic event identifier and sort the alternative path set; Extract the event severity score, judge whether the path switching condition is satisfied, if the path switching condition is satisfied, extract the highest priority path from the alternative path set to generate an updated recommended path.

[0009] In a second aspect, the present application provides an intelligent guidance system, the system includes: A user preference feature vector generation module, which obtains user historical behavior data from the navigation system database and generates a user preference feature vector based on the user historical behavior data; A real-time traffic status parameter matrix generation module, which sets a personalized traffic flow threshold according to the user preference feature vector, starts a real-time monitoring mechanism, collects current traffic status data, and generates a real-time traffic status parameter matrix; An optimal weight configuration scheme generation module, which dynamically adjusts the time cost weight, distance cost weight, and congestion cost weight in path planning for the real-time traffic status parameter matrix and the user preference feature vector by using the gradient descent algorithm to generate an optimal weight configuration scheme; A path priority sorted list generation module, which obtains road network topology data, generates a set of candidate paths, comprehensively scores the set of candidate paths according to the optimal weight configuration, and generates a path priority sorted list; The final optimized navigation path scheme generation module monitors the real-time traffic status of the recommended path according to the path priority sorted list. If an anomaly is detected, it generates an alternative path set and calculates the path switching timing judgment condition, and performs dynamic path adjustment according to the path switching timing judgment condition to generate the final optimized navigation path scheme.

[0010] In a third aspect, the present application provides an intelligent guidance device, which includes: a memory, a processor, and an intelligent guidance program stored on the memory and executable on the processor. The intelligent guidance program is configured to implement the intelligent guidance method as described above.

[0011] In a fourth aspect, the present application provides a computer-readable storage medium, on which an intelligent guidance program is stored. When the intelligent guidance program is executed by a processor, it implements the intelligent guidance method as described above.

[0012] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: The present invention discloses an intelligent navigation method based on user preferences. By analyzing historical user behavior data, a preference feature vector is generated, a personalized traffic flow threshold is set, and the road conditions are monitored in real time. When the threshold is exceeded, a dynamic analysis module is triggered to generate a traffic state parameter matrix. Combining user preferences and real-time road conditions, the present invention uses the gradient descent algorithm to dynamically adjust the path planning weights, constructs an optimal weight configuration scheme, comprehensively scores and sorts the candidate paths. The system continuously monitors the status of the recommended path, generates alternative paths and judges the switching timing when necessary, and obtains the final output optimized navigation scheme. By analyzing historical behavior data to generate feature vectors, the path planning fits the individual habits of users; by combining user preferences with real-time traffic parameters and dynamically adjusting the path weights, the "one-size-fits-all" type of recommendation of traditional navigation is avoided; by continuously monitoring traffic changes, when a preset condition (such as congestion deterioration exceeding the threshold) is triggered, the candidate paths are automatically re-evaluated, reducing the need for manual adjustment by users. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 is a specific flowchart of an intelligent guidance method of the present invention; Figure 2 is a specific flowchart of step S103 of the intelligent guidance method in this embodiment; Figure 3 is a schematic structural diagram of an intelligent guidance system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0015] AsFigure 1 , a smart guidance method in this embodiment may specifically include: S101. Obtain user historical behavior data from the navigation system database, and generate a user preference feature vector based on the user historical behavior data; In this embodiment, user historical behavior data is obtained. The user historical behavior data includes travel time distribution, route selection records, and speed change sequences. Multidimensional features are extracted through principal component analysis and clustering algorithms to generate a user preference feature vector, which includes a time preference component, a distance preference component, and a congestion preference component.

[0016] Further, extract user historical behavior data from the navigation system database. The data covers travel time distribution and route selection records, construct an initial data set, and obtain a multidimensional information set of user behavior. According to the initial data set, standardize the multidimensional information of user behavior to eliminate the dimension difference between different dimension data, and obtain a standardized data matrix. Use the principal component analysis method to perform dimensionality reduction on the standardized data matrix, extract the main feature components, and obtain a dimensionality-reduced feature data set. Group analysis is performed on the dimensionality-reduced feature data set through a clustering algorithm to divide the categories of time preference, distance preference, and congestion preference, and obtain the classification result of user behavior. According to the classification result, calculate the core eigenvalue of user behavior in each category, and construct a feature vector including a time preference component, a distance preference component, and a congestion preference component to obtain a user preference feature vector.

[0017] Specifically, extract user historical behavior data from the navigation system database, use SQL query statements to filter travel records in the past 6 months, obtain travel time distribution data such as the proportion of morning rush hour is 35% and the proportion of evening rush hour is 28%, the route selection records include the path ID and frequency selected by the user, and the speed change sequence records the speed change by sampling once per second. Perform Z-score standardization processing on the initial data set to normalize dimensions such as travel time, route frequency, and average speed to the interval [-1, 1]. Use the PCA principal component analysis algorithm to set the variance contribution rate threshold at 85%, extract the first 3 principal components as feature components, and generate a dimensionality-reduced data set with a cumulative variance of 92%. Use the K-means clustering algorithm to set the number of clusters k = 3, iteratively calculate the cluster centers based on the Euclidean distance, and divide the categories of time preference (such as concentrated commuting periods), distance preference (such as the proportion of short-distance first is 60%), and congestion preference (such as the probability of avoiding congestion is 0.2). Calculate the center point coordinates of each cluster according to the clustering result, and construct user preference feature vectors corresponding to time preference, distance preference, and user preference respectively.

[0018] S102. Set a personalized traffic flow threshold according to the user preference feature vector, start the real-time monitoring mechanism, collect the current traffic state data, and generate a real-time traffic state parameter matrix. In this embodiment, according to the user preference feature vector, a personalized traffic flow threshold is set, the real-time monitoring mechanism is started, and the current traffic flow, road congestion index, and weather environment data are continuously collected. If the traffic flow exceeds the preset threshold, the dynamic analysis module is triggered to generate a real-time traffic state parameter matrix, and the real-time traffic state parameter matrix includes a road passing speed component, a congestion index component, and a weather impact component.

[0019] Further, according to the user preference feature vector, by analyzing the historical behavior data and personalized needs, a personalized traffic flow threshold is set to obtain a user-specific traffic limit standard. Through the set personalized traffic flow threshold, the real-time monitoring mechanism is started, and the current traffic flow, road congestion index, and weather environment data are continuously collected from the sensors and meteorological interfaces to obtain a real-time collection data set. According to the real-time collection data set, data cleaning and standardization processing methods are used to remove outliers and unify the data format to determine the processed structured traffic data. If the traffic flow in the processed structured traffic data exceeds the personalized traffic flow threshold, the dynamic analysis module is triggered to start the real-time traffic state evaluation process to obtain a preliminary evaluation result. According to the preliminary evaluation result, by calling the multi-dimensional analysis algorithm in the dynamic analysis module, the road passing speed component, the congestion index component, and the weather impact component are calculated to generate a real-time traffic state parameter matrix. If any component in the real-time traffic state parameter matrix exceeds the preset range, through the data fusion technology, the matrix data is compared with the historical traffic pattern to judge the abnormal degree of the current traffic state. According to the abnormal degree judgment result, a weight adjustment algorithm is used to optimize the real-time traffic state parameter matrix, update the influence ratio of each component in the matrix, and obtain an optimized state parameter matrix. Through the optimized state parameter matrix, combined with the weather environment data in the real-time collection data set, a secondary verification algorithm is executed to determine the real-time traffic state parameter matrix.

[0020] Specifically, according to the user preference feature vector [time preference, distance preference, congestion preference], the K-means clustering algorithm is used to analyze the user's historical travel data (such as peak period selection, distance preference) to obtain the user tolerance cluster. Specifically, the user tolerance cluster is obtained through the following formula: , represents the optimal clustering result of the K-means clustering algorithm, k represents the number of clustering clusters, represents the i-th clustering cluster, x represents the user preference feature vector including time preference, distance preference, and congestion preference, represents the center point of the $i$-th cluster. This formula realizes the optimal clustering division of user tolerance by minimizing the sum of squared errors within the cluster to obtain the user tolerance clusters.

[0021] Based on the user tolerance clusters, set personalized traffic flow thresholds for users, which can refer to the following formula: , represents the personalized basic traffic flow threshold, represents the initial flow threshold, $\alpha$ represents the adjustment coefficient, represents the weight factor of the tolerance cluster to which the user belongs. This formula dynamically adjusts the personalized traffic flow threshold according to the characteristics of the user tolerance cluster.

[0022] Specifically: The obtained historical user behavior shows that 70% of trips are during the evening rush hour (17:00 - 19:00), and the user prefers the "shortest distance" route. In the past month, when the current traffic flow > 750 vehicles per hour, the probability that user B takes a detour actively reaches 80%. Through K-means clustering, user B is assigned to the medium tolerance cluster, and at this time, the personalized basic threshold is obtained as 800 vehicles per hour. The vehicle flow data is collected by microwave sensors deployed on the roadside at a sampling rate of 1 Hz. At the same time, the API of the meteorological bureau is called to obtain the precipitation intensity (1.2 mm / h), visibility (500 m), and wind speed (5 m / s), and a triple dataset including the current traffic flow, congestion index (0 - 10 levels), and weather is constructed. Among them, the congestion index reflects the degree of deterioration of the current traffic state relative to the free flow (unobstructed) state, and is commonly expressed in 0 - 10 levels or percentages. The general calculation model of the congestion index is: congestion index = (1 - free flow speed / current average speed) × scaling factor + additional correction term. The free flow speed refers to the road design speed or the average speed during historical congestion-free periods. The scaling factor: maps the result to the target range (e.g., multiply by 10 for 0 - 10 levels); additional correction term: considers temporary factors such as weather and accidents (e.g., +1 level during heavy rain). Based on the above formula, the congestion index is obtained as 7.8. The Z-score normalization method is used to process the original data, and outliers deviating from the mean ± 3σ are removed (such as the sudden drop in traffic flow to 0 caused by GPS drift), and structured JSON format data is output. When the traffic flow of a certain section is detected to reach 850 vehicles per hour, the gradient boosting decision tree model of the dynamic analysis module is activated, and the input features are: current traffic flow (850), congestion index (7.8), and precipitation intensity (1.2 mm / h). Calculate the passing speed component (km / h) = basic speed (such as 80 Km / h) × (1 - 0.2 × congestion coefficient) × (1 - 0.15 × precipitation intensity) = 65 m / h. The congestion index component 7.8 in the matrix exceeds the threshold 7.5, so it is matched with the data of the same time period in the past 30 days using the DTW algorithm, and the anomaly score is output. Based on the anomaly score, the entropy weight method is used to reallocate the matrix weights, and the weather impact component is adjusted from 0.3 to 0.4. Combining the current wind speed (≥8 m / s) data, the passing speed prediction value is corrected through Kalman filtering, and finally a real-time traffic state parameter matrix including the road passing speed component, congestion index component, and weather impact component of [65 km / h, 7.8, 0.4] is output.

[0023] S103. For the real-time traffic state parameter matrix and the user preference feature vector, the gradient descent algorithm is used to dynamically adjust the time cost weight, distance cost weight, and congestion cost weight in the path planning to generate an optimal weight configuration scheme.

[0024] In this embodiment, as Figure 2As shown in the figure, it includes: extracting data related to the time cost, distance cost, and congestion cost of the current road based on the real-time traffic state parameter matrix to obtain the initial cost distribution characteristics; Determining a comprehensive cost calculation model by combining the initial cost distribution characteristics with the user preference feature vector; using the gradient descent algorithm to dynamically adjust the adjustment range of the time cost weight, distance cost weight, and congestion cost weight to generate a preliminary weight configuration combination; if the comprehensive cost of the preliminary weight configuration combination exceeds the preset threshold, then update the preliminary weight configuration combination to generate an optimal weight configuration plan.

[0025] Furthermore, based on the real-time traffic state parameter matrix, extract data related to the time cost, distance cost, and congestion cost of the current road to obtain the initial cost distribution characteristics. Through the initial cost distribution characteristics, combined with the priority ranking in the user preference feature vector, construct a cost function for path planning to determine the comprehensive cost calculation model. Use the gradient descent algorithm to iteratively optimize the comprehensive cost calculation model, calculate the adjustment direction of the time cost weight, distance cost weight, and congestion cost weight, and obtain the preliminary weight adjustment range. Among them, the comprehensive cost calculation model is: , C represents the value of the comprehensive cost function, represents the time cost weight, T represents the time cost, represents the distance cost weight, D represents the distance cost, represents the congestion cost weight, G represents the congestion cost. This formula defines a comprehensive cost calculation model including three cost elements. According to the preliminary weight adjustment range, update the values of the time cost weight, distance cost weight, and congestion cost weight to generate a temporary weight configuration combination. Through the temporary weight configuration combination, recalculate the comprehensive cost value in path planning to obtain the updated cost distribution result. If the updated cost distribution result does not meet the preset convergence condition, then return to the gradient descent algorithm for a new round of iterative optimization to determine whether further adjustment of the weight values is required. If the updated cost distribution result meets the preset convergence condition, then solidify the temporary weight configuration combination into a candidate weight configuration plan to determine a preliminary plan including the time weight value, distance weight value, and congestion weight value. According to the candidate weight configuration plan, combined with the dynamic change data in the real-time traffic state parameter matrix, conduct final verification and fine-tuning to obtain the optimal weight configuration plan. Through the optimal weight configuration plan, output the final path planning parameters including the time weight value, distance weight value, and congestion weight value to complete the dynamic adjustment process.

[0026] Specifically, based on the time cost (e.g., the travel time of a certain section is 12 minutes = distance cost / road travel speed component = 13 / 65), distance cost (e.g., 13 kilometers), and congestion cost (e.g., congestion index 0.78) of the current road obtained from the real-time traffic state parameter matrix, where the congestion cost is positively correlated with the weather impact component, combined with the priority ranking in the user preference feature vector (e.g., time weight 0.6, distance weight 0.3, congestion weight 0.1), an initial cost distribution feature is formed. Based on this feature, a comprehensive cost calculation model is constructed, using the linear weighting formula: total cost = time weight × time cost + distance weight × distance cost + congestion weight × congestion cost = 0.6 * 12 + 13 * 0.3 + 0.78 * 0.1 = 11.2. The gradient descent algorithm is used, specifically using the following formula: , where, represents the updated weight, represents the current weight, and α represents the learning rate. represents the gradient of the cost function C with respect to the weight w. This formula describes the iterative update rule of the weight parameters in the gradient descent algorithm. For example, setting the learning rate to 0.1, calculating the weight adjustment direction (e.g., time weight gradient 0.2, distance weight gradient -0.1, congestion weight gradient -0.05), generating a preliminary adjustment amplitude. Update the weight values to form a temporary weight combination (0.58, 0.31, 0.105). Recalculate the comprehensive cost value (e.g., the comprehensive cost drops from 11.2 to 9.9). If the convergence threshold is not reached (e.g., cost change > 0.1), continue iterative optimization. When the comprehensive cost change ≤ 0.1, output the final weight configuration of time, distance, and congestion as (0.47, 0.32, 0.13) for path planning.

[0027] S104. Obtain the road network topology structure data, generate a set of candidate paths, and comprehensively score the set of candidate paths according to the optimal weight configuration to generate a path priority ranking list.

[0028] In this embodiment, extract the road network topology structure data and node connection relationship data from a preset database, construct an initial path network model; according to the initial path network model, use a graph traversal algorithm to generate a set of candidate paths; comprehensively score the set of candidate paths according to the optimal weight configuration to generate a path priority ranking list.

[0029] Specifically, assume there is a simple urban road network, including the following nodes and road segments: Nodes: A, B, C, D, E, F, G, H (representing intersections or landmarks); Road segments: roads connecting each node, and each road has attributes such as distance, speed limit, and road type. It is necessary to plan the optimal path from the starting point A to the ending point H, including: obtaining the topological structure data of the road network and extracting the road network data from the database; generating a set of candidate paths, using the A* algorithm to generate candidate paths from the starting point A to the ending point H. Among them, the A* algorithm is a widely used heuristic search algorithm for finding the optimal path from the starting point to the ending point in a graph or network. Its core idea is to evaluate each possible moving direction and select the path with the smallest combined cost for expansion. For example, (g(n)) is the actual cost (known) from the starting point to the current node (n), (h(n)) is the estimated cost (heuristic function) from the current node (n) to the ending point, and (f(n) = g(n) + h(n)) is the total cost used to determine which node to expand to next. Then the ultimate goal is to determine which node to expand to next under the condition of the smallest total cost.

[0030] In this example, assume the following candidate paths are found: 1. A -> B -> C -> F -> H 2. A -> B -> E -> F -> H 3. A -> D -> E -> F -> H 4. A -> D -> G -> H 5. A -> B -> E -> G -> H Conduct a comprehensive scoring of the candidate paths according to the optimal weight configuration. The final weight configuration for time, distance, and congestion is (0.47, 0.32, 0.13), and these weights reflect the importance that users attach to different factors. Generate a sorted list of path priorities and sort the paths according to the scoring results: A -> B -> E -> F -> H, distance: 20 km, estimated time: 22 minutes, number of congestion times: 3, comprehensive score: 0.47 * 22 + 0.32 * 20 + 0.13 * 3 = 17.13; A -> B -> C -> F -> H, distance: 20 km, estimated time: 24 minutes, number of congestion times: 2, comprehensive score: 0.47 * 24 + 0.32 * 20 + 0.13 * 2 = 17.94; A -> D -> E -> F -> H, distance: 15 km, estimated time: 27 minutes, number of congestion times: 5, Comprehensive score: 0.47 * 27 + 0.32 * 15 + 0.13 * 5 = 18.14; A -> B -> E -> G -> H, Distance: 18 km, Estimated time: 29 minutes, Number of congestions: 2, Comprehensive score: 0.47 * 29 + 0.32 * 18 + 0.13 * 2 = 19.65; A -> D -> G -> H, Distance: 15 km, Estimated time: 30 minutes, Number of congestions: 7, Comprehensive score: 0.47 * 30 + 0.32 * 15 + 0.13 * 7 = 19.81; Based on the above comprehensive score ranking, a path priority ranking list <4, 5, 3, 1, 2> is obtained.

[0031] S105. Monitor the real-time traffic status of the recommended path according to the path priority ranking list. If an abnormality is detected, generate an alternative path set and calculate the path switching timing judgment condition. Perform dynamic path adjustment according to the path switching timing judgment condition to generate a final optimized navigation path plan; In this embodiment, for the path priority ranking list and the traffic condition change trend, monitor the real-time traffic status of the recommended path. If a sudden traffic event or an increase in the congestion probability value is detected through the real-time traffic status parameter matrix, generate an alternative path set, and calculate the path switching timing judgment condition based on the event severity component and the influence range component.

[0032] Furthermore, obtain the real-time traffic status data, extract the real-time traffic status parameter matrix, and analyze the traffic condition change trend; if the traffic condition change trend shows abnormal fluctuations, determine that there is a sudden traffic event; if the congestion probability value exceeds the preset threshold, generate a sudden traffic event identifier; extract the alternative path set according to the sudden traffic event identifier, and sort the alternative path set; extract the event severity score, and judge whether the path switching condition is met. If the path switching condition is met, extract the highest priority path from the alternative path set to generate an updated recommended path.

[0033] Specifically, assume that the initial recommended route is Route 4: A -> D -> G -> H. When the traffic monitoring system is running, the following situations will occur: Real-time traffic data acquisition: The system acquires traffic data at the current time (8:00 am); it is found that there are sudden traffic incidents on the road D -> G and on the road BE, and the congestion index rises from 0.5 to 0.9, with an increase of 0.4 which is greater than the preset threshold of 0.2. Then a sudden traffic incident identifier is generated. According to the sudden traffic incident identifier, the set of alternative routes is extracted, and the set of alternative routes is sorted as <5, 3, 1, 2>; in addition, the estimated time changes from the original 30 minutes to 45 minutes, and the number of congestion times changes from the original 7 times to 10 times; the system analyzes the trend of road condition changes, identifies the abnormal fluctuations of Route 4: A -> D -> G -> H and generates a sudden traffic incident identifier, and the event severity score is: 0.47 * (45 - 30) + 0.13 * (10 - 7) = 7.44; Route switching timing judgment: The switching timing is: whether the score obtained by subtracting the event severity score from the initial recommended route is still greater than the score of the highest-ranked alternative route. Here, the score of the initial recommended route 19.81 minus 7.44 is 12.37 which is less than the total score of Route 5 in the alternative route priority ranking. Therefore, at this time, a route switch is made, and the final optimized navigation route obtained is Route 5: A -> B -> E -> G -> H.

[0034] As Figure 3 shown, an embodiment of the present invention further provides an intelligent guidance system for implementing the above-mentioned intelligent guidance method. The system includes multiple modules that work together to optimize the operating parameters of intelligent guidance. The system includes: A user preference feature vector generation module that obtains user historical behavior data from the navigation system database and generates a user preference feature vector based on the user historical behavior data; A real-time traffic state parameter matrix generation module that sets personalized traffic flow thresholds according to the user preference feature vector, starts a real-time monitoring mechanism, collects current traffic state data, and generates a real-time traffic state parameter matrix; An optimal weight configuration scheme generation module that, for the real-time traffic state parameter matrix and the user preference feature vector, uses the gradient descent algorithm to dynamically adjust the time cost weight, distance cost weight, and congestion cost weight in route planning to generate an optimal weight configuration scheme; A route priority ranking list generation module that obtains road network topology structure data, generates a set of candidate routes, and comprehensively scores the set of candidate routes according to the optimal weight configuration to generate a route priority ranking list; The final optimized navigation path plan generation module monitors the real-time traffic status of the recommended path according to the path priority sorting list. If an abnormality is detected, it generates an alternative path set and calculates the path switching timing judgment condition, and performs dynamic path adjustment according to the path switching timing judgment condition to generate the final optimized navigation path plan.

[0035] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be combined arbitrarily with each other.

Claims

1. An intelligent guidance method, characterized in that, The method includes: obtaining user historical behavior data from a navigation system database, generating a user preference feature vector based on the user historical behavior data; setting a personalized traffic flow threshold according to the user preference feature vector, starting a real-time monitoring mechanism, collecting current traffic state data, and generating a real-time traffic state parameter matrix; for the real-time traffic state parameter matrix and the user preference feature vector, using a gradient descent algorithm to dynamically adjust the time cost weight, distance cost weight, and congestion cost weight in path planning, generating an optimal weight configuration scheme; obtaining road network topology structure data, generating a candidate path set, comprehensively scoring the candidate path set according to the optimal weight configuration, and generating a path priority ranking list; monitoring the real-time traffic state of the recommended path according to the path priority ranking list, if an anomaly is detected, generating an alternative path set and calculating a path switching timing judgment condition, and performing dynamic path adjustment according to the path switching timing judgment condition to generate a final optimized navigation path scheme.

2. The intelligent guidance method according to claim 1, wherein The obtaining user historical behavior data from a navigation system database and generating a user preference feature vector includes: extracting user historical behavior data from the navigation system database, where the user historical behavior data includes travel time distribution and route selection records; performing dimensionality reduction processing on the user historical behavior data, extracting main feature components, and determining the key patterns of user behavior; grouping user behavior according to the key patterns, dividing time preference, distance preference, and congestion preference; constructing a user preference feature vector according to the time preference component, the distance preference component, and the congestion preference.

3. The intelligent guidance method according to claim 1, wherein, The setting a personalized traffic flow threshold according to the user preference feature vector, starting a real-time monitoring mechanism, collecting current traffic state data, and generating a real-time traffic state parameter matrix includes: classifying users through the user preference feature vector to determine the personalized traffic flow threshold; if the personalized traffic flow threshold is determined, starting the real-time monitoring mechanism, collecting current traffic flow, road congestion index, and weather environment data, and constructing a real-time acquisition data set; extracting the traffic flow change trend, congestion index fluctuation characteristics, and weather impact weight from the real-time acquisition data set; if the traffic flow component exceeds the personalized traffic flow threshold, triggering a dynamic analysis module to generate a real-time traffic state parameter matrix including a passing speed component, a congestion index component, and a weather impact component.

4. The intelligent guidance method according to claim 1, wherein The using a gradient descent algorithm to dynamically adjust the time cost weight, distance cost weight, and congestion cost weight in path planning for the real-time traffic state parameter matrix and the user preference feature vector to generate an optimal weight configuration scheme includes: Based on the real-time traffic state parameter matrix, obtaining data related to the time cost, distance cost, and congestion cost of the current road to obtain an initial cost distribution feature; Determine a comprehensive cost calculation model by combining the initial cost distribution characteristics with the user preference feature vector; use the gradient descent algorithm to dynamically adjust the adjustment amplitudes of the calculation time cost weight, distance cost weight, and congestion cost weight to generate a preliminary weight configuration combination; If the comprehensive cost of the preliminary weight configuration combination exceeds the preset threshold, update the preliminary weight configuration combination to generate an optimal weight configuration scheme.

5. The intelligent guidance method according to claim 1, characterized in that The obtaining of the road network topology structure data, generating a candidate path set, and comprehensively scoring the candidate path set according to the optimal weight configuration to generate a path priority sorted list includes: extracting the road network topology structure data and node connection relationship data from a preset database to construct an initial path network model; Generate a candidate path set by using a graph traversal algorithm according to the initial path network model; Comprehensively score the candidate path set according to the optimal weight configuration to generate a path priority sorted list.

6. The intelligent guidance method according to claim 1, characterized in that, Monitoring the real-time traffic status of the recommended path according to the path priority sorted list, if an abnormality is detected, generating an alternative path set and calculating the path switching opportunity judgment condition, and performing dynamic path adjustment according to the path switching opportunity judgment condition to generate a final optimized navigation path scheme, including: obtaining the real-time traffic status data, extracting the real-time traffic status parameter matrix, and analyzing the traffic condition change trend; If the traffic condition change trend shows abnormal fluctuations, determine that there is a sudden traffic event; if the congestion probability value exceeds the preset threshold, generate a sudden traffic event identifier; Extract the alternative path set according to the sudden traffic event identifier and sort the alternative path set; Extract the event severity score, judge whether the path switching condition is satisfied, and if the path switching condition is satisfied, extract the highest priority path from the alternative path set to generate an updated recommended path as the optimal path.

7. An intelligent guiding system, characterized in that, The system includes: The user preference feature vector generation module obtains the user's historical behavior data from the navigation system database and generates a user preference feature vector based on the user's historical behavior data; the real-time traffic state parameter matrix generation module sets a personalized traffic flow threshold according to the user preference feature vector, starts a real-time monitoring mechanism, collects current traffic state data, and generates a real-time traffic state parameter matrix; the optimal weight configuration scheme generation module dynamically adjusts the time cost weight, distance cost weight, and congestion cost weight in path planning by using the gradient descent algorithm for the real-time traffic state parameter matrix and the user preference feature vector, and generates an optimal weight configuration scheme; the path priority sorting list generation module obtains the road network topology structure data, generates a candidate path set, comprehensively scores the candidate path set according to the optimal weight configuration, and generates a path priority sorting list; the final optimized navigation path scheme generation module monitors the real-time traffic state of the recommended path according to the path priority sorting list, if an abnormality is detected, generates an alternative path set and calculates the path switching opportunity judgment condition, and performs dynamic path adjustment according to the path switching opportunity judgment condition to generate a final optimized navigation path scheme.

8. An intelligent guiding device, characterized in that, The intelligent guidance device includes: a memory, a processor, and an intelligent guidance program stored on the memory and executable on the processor, and the intelligent guidance program is configured to implement the intelligent guidance method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, An intelligent guidance program is stored on the computer-readable storage medium, and when the intelligent guidance program is executed by a processor, it implements the intelligent guidance method according to any one of claims 1 to 6.

Images