A method for optimizing emergency vehicle routing by integrating meteorological data and vehicle parameters

Abstract

The present invention discloses a method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters. The method comprises the following steps: S1, constructing a multidimensional database; S2, improving the route optimization algorithm, including dynamic weight allocation; multi-objective optimization; S3, dynamic route adjustment: deploying a lightweight model on the vehicle terminal to receive meteorological and vehicle data in real time and update the optimal route; S4, human-machine collaborative intervention: when the vehicle deviates from the planned route or stays above a threshold, route replanning is initiated, and alternative routes are generated in combination with manual adjustments by the driver. This method integrates meteorological data and vehicle parameters to construct a multidimensional database. A weather-vehicle parameter coupling factor is introduced into route planning to perform dynamic weight allocation; total travel time, weather risk, and vehicle loss are incorporated into the objective function to perform multi-objective optimization; and a model is trained using historical data to predict the optimal route under different weather-vehicle combinations.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle control and regulation systems, and in particular to a method for optimizing a repair vehicle path by integrating meteorological data and vehicle parameters. Background Art

[0002] Optimizing repair vehicle routes is a critical component of the emergency response system, playing an irreplaceable role in improving rescue efficiency and minimizing social losses. In emergency response, repair tasks typically need to be completed as quickly as possible to ensure the normal operation of systems or equipment. Scientifically planned routes can reduce average response times by 30%-50%. Regarding resource utilization, traditional empirical scheduling often results in vehicles taking detours, running empty, or overloaded, exacerbating waste of manpower and material resources. Scientifically planned routes, integrating information such as task point distribution and road conditions into algorithmic models, can achieve balanced vehicle loads and minimize mileage, significantly improving the efficiency of equipment and personnel scheduling. Regarding safety risks, severe weather (such as heavy rain, ice, and snow) can significantly increase traffic accident rates. Scientifically planned routes can dynamically avoid high-risk road sections, adjust driving strategies, and mitigate operational risks in extreme environments, thus ensuring a safe and secure response for repair missions.

[0003] The following related patents were found after searching:

[0004] 1. Chinese patent CN117537836A discloses a navigation path optimization method that considers real-time accident risks and proposes a risk-based navigation path optimization algorithm that can more accurately predict path risks.

[0005] 2. Chinese patent CN107843252A discloses a navigation path optimization method, device, and electronic device. By screening the node set corresponding to the navigation path to be optimized, the number of turns in the path to be optimized can be reduced, thereby optimizing the navigation route.

[0006] The problem with the above patent is that existing path optimization research focuses on a single data source, and focuses on weather, accidents and task point distribution, while ignoring parameters such as vehicle driving mode, tire type, vehicle performance, load, and endurance. There are still significant deficiencies in the integrated application of weather data and vehicle parameters, which makes it difficult for path planning to accurately adapt to complex environments. Summary of the Invention

[0007] The present invention aims to address at least one of the technical problems existing in the prior art. To this end, one objective of the present invention is to propose a method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters. This method integrates meteorological data and vehicle parameters to construct a multidimensional database. Path factors are introduced into route planning to dynamically assign weights. Total travel time, weather risk, and vehicle losses are incorporated into the objective function for multi-objective optimization. A model is trained using historical data to predict the optimal route under different weather-vehicle combinations.

[0008] In order to solve the above problems, the present invention provides a method for optimizing the repair vehicle path by integrating meteorological data and vehicle parameters, comprising the following steps:

[0009] S1. Build a multi-dimensional database: Integrate real-time and historical weather data to extract macro-weather indicators and micro-road environment indicators; establish a vehicle performance database that includes drive mode, tire type, load capacity, and range;

[0010] S2. Improved path optimization algorithm, including:

[0011] Dynamic weight allocation: Calculates the path factor by integrating weather coefficients including macro and micro weather indicators and vehicle performance parameters;

[0012] Multi-objective optimization: Construct an objective function with total travel time, weather risk, and vehicle loss as variables, and solve the multi-objective problem using an improved genetic algorithm;

[0013] S3, Dynamic route adjustment: Deploy a lightweight model on the vehicle terminal to receive real-time weather and vehicle data and update the optimal route;

[0014] S4, Human-machine collaborative intervention: When the vehicle deviates from the planned path or stays for more than a threshold, path replanning is initiated and an alternative path is generated in combination with the driver's manual adjustments.

[0015] Preferably, the specific steps of constructing the multidimensional database in S1 are as follows:

[0016] S101. Build a meteorological database. Through standardized APIs, obtain minute-by-minute meteorological data in real time, covering basic weather indicators based on latitude and longitude, calculate micro-weather indicators, analyze meteorological warning information, and collect historical data from public meteorological databases, including hourly extreme weather events, corresponding road congestion duration, historical accident rates, and vehicle dispatch data from the vehicle management platform.

[0017] S102. Data cleaning and standardization: outliers are removed, time formats are unified, and spatial coordinates are converted to WGS84 latitude and longitude. A hierarchical storage architecture is designed, using a time series database to store real-time data streams and support second-level queries. Macro and micro indicators are extracted through ETL tools and stored in a relational database. Spatiotemporal indexes are established to accelerate regional weather queries.

[0018] S103: Constructing a vehicle parameter database. Basic attributes such as vehicle drive mode, tire type, load, and range are obtained from the vehicle management platform's resource management module and encoded into a standardized dictionary. Real-time load and other data are obtained from the vehicle management platform's dispatch order data. This data is combined with remaining battery life, tire pressure, and brake system information to construct a dynamic vehicle parameter profile. Average vehicle performance parameters under various weather conditions are obtained through test sites to form a vehicle environmental adaptability matrix. Finally, a weather-vehicle parameter mapping table, a range attenuation model, and a load constraint library are established.

[0019] S104. Using "path ID + timestamp + vehicle ID" as the primary key, a real-time association table of weather data and vehicle parameters is established; a standardized calling interface is provided to support real-time acquisition of path factors during path optimization algorithms.

[0020] Preferably, the algorithm and constraints need to be improved in S2. The specific steps are as follows:

[0021] S201. Dynamic weight allocation: Introduce path factors in path planning; sort out the adverse weather conditions that affect vehicle travel and divide them into 15 types, namely rainstorm, thunderstorm, heavy rainfall, typhoon, hurricane, tornado, strong wind, high temperature, hail, freezing, freezing, blizzard, heavy snow, haze, and sandstorm; path factor J i = path distance × ;W fa is the weather macro-index coefficient, with a value of 0.3~0.6; W fb is the weather micro-index coefficient, with a value of 0.3~0.6; V i is the vehicle parameter coefficient, assigned a value of 0.3~0.7; U t is the path coefficient, which defaults to 1. When the corresponding path of the vehicle is risky, the path coefficient will be increased;

[0022] S202, Multi-Objective Optimization: Incorporate total driving time, weather risk, and vehicle loss into the objective function; through an improved genetic algorithm, balance the two objectives of "shortest time" and "minimizing vehicle loss in extreme weather" to improve route planning accuracy;

[0023] The formula of the initial scheduling model for path optimization is as follows:

[0024] ;

[0025] Where, X=(x1,x2,...,x m ), x1,x2,...,x m , m is the vehicle loss dimension, f1(X) is the minimum variable for total repair time, and f2(X) is the minimum variable for vehicle loss under extreme weather conditions;

[0026] S203. Based on the massive amount of historical repair task data in the vehicle management platform, including weather data and vehicle parameter data, data governance is performed according to the time dimension. The data is cleaned and normalized, key feature information is extracted, and a high-dimensional feature vector is constructed as the model input. A deep network architecture is adopted, with path factors as the input layer and path decisions as the output layer. The state space is defined to include the current location, destination, real-time weather, and vehicle parameters; the behavior space is the optional path branch; the reward function is designed as the path optimization initial scheduling model; the preprocessed historical data is divided into a training set E = (e1, e2, ..., e m )、Validation set V=(v1,v2,...,v m ), and the test set T=(t1,t2,...,t m During training, the agent selects a path based on its current state in a simulated repair scenario, obtains reward feedback through interaction with the environment, continuously optimizes its strategy, monitors model performance using the validation set V, adjusts hyperparameters to prevent overfitting, and finally converges with the test set T to evaluate its ability to predict the optimal path.

[0027] Preferably, the calculation formula of the minimum total path time variable f1(X) in S202 is as follows:

[0028] ;

[0029] Where P is the dispatch task, D is the corresponding route selection in each dispatch task, R is the path selection on the road, and J is the route selection on the road. i is the path factor, T i is the average time spent on the i-th path under normal weather conditions, J wk is the path waiting time, X vg As the constraint condition for repairing vehicles, each dispatch task is assigned at least one vehicle. Under the condition of no human intervention and normal road conditions, X vg =1;

[0030] The calculation formula for the minimum vehicle loss variable f2(X) under the influence of extreme weather in S202 is as follows:

[0031] ;

[0032] Where, J i is the path factor, U iis the vehicle loss caused by the adverse weather factors on the i-th path, L i is the loss accumulation factor of the i-th path.

[0033] Preferably, the path optimization initial scheduling model in S202 has two pre-constraints:

[0034] Normal driving capability of emergency vehicles under extreme weather conditions Y i is an integer greater than zero, Y i This is the vehicle inspection data in the vehicle management platform. This is constraint condition 1. The formula is as follows:

[0035] Y i ≥0;

[0036] Vehicle loss caused by extreme weather i Less than the delay loss H of the vehicle dispatch order i For constraint 2, the formula is as follows:

[0037] ;

[0038]

[0039]

[0040] Where α is the impact ratio of extreme weather;

[0041] On the premise that the above two constraints are met, the calculation of the path optimization initial scheduling model in S202 is performed.

[0042] Preferably, in S202, the path optimization parameter variables are optimized using an improved genetic algorithm, and the optimization content includes:

[0043] A two-layer coding mode is used to encode the path optimization parameters in the genetic algorithm. The first layer of the path layer adopts integer coding, and the second layer of the path factor is converted into binary coding.

[0044] Dynamically adjust weights through genetic operations to achieve adaptive allocation of path factors;

[0045] The mutation of each binary bit can fine-tune the parameters, and crossover can combine the advantages of different parameter combinations;

[0046] Set the initial parameters related to path optimization, including the initial path, iterations, and search times;

[0047] Determine the path optimization fitness function as the inverse of the objective function to ensure maximum fitness;

[0048] Determine the strategy for selecting the path and the criteria for stopping iteration, then perform crossover and mutation operations. When the set number of iterations is reached, determine the target parameter variable.

[0049] Preferably, the selection strategy and stopping criteria in S202 are expressed as follows:

[0050] ;

[0051] ;

[0052] Where, W(x i ) is the probability of each individual being inherited to the next generation, U(x i ) is the probability of chromosome being selected.

[0053] Preferably, S3 needs to be based on edge computing. During emergency repair tasks, the on-board terminal collects weather and vehicle data in real time, inputs the trained path optimization scheduling model, supports the rapid output of the optimal path, and the driver drives according to the optimal path, and continuously feeds back data to dynamically optimize the model.

[0054] Preferably, in S4, in the human-machine collaborative mode, the driver is allowed to manually intervene in the route according to the actual road conditions; when the vehicle deviates from the optimal planned path by more than 500 meters or stops for more than 3 minutes, the system automatically prompts "Do you want to adjust the route due to road conditions?" and the driver confirms and starts re-planning.

[0055] Preferably, in S4, in the human-machine collaborative mode, the driver intervention instruction is converted into a dynamic emergency vehicle constraint condition X vg , and integrate with real-time meteorological data and vehicle parameters to form a new state space; using an improved genetic algorithm, taking the current vehicle position as a new starting point, searching for feasible paths within the local road network, while retaining the effective sections that have not been interfered with in the original plan, the calculation time is controlled within 5 seconds; based on the path factor J i The path coefficient U involved in t , dynamically adjust the weights, give priority to meeting the driver's intervention intention, and balance the two goals of "shortest time" and "minimum vehicle loss in extreme weather", and then generate alternative paths for the driver to confirm again.

[0056] The advantages of the present invention compared with the prior art are:

[0057] 1. Accurately respond to complex environments and improve path planning reliability: This invention constructs a dynamic coupling model by quantitatively analyzing vehicle drive mode, tire type, load capacity, and road slipperiness, visibility, wind speed, and other parameters. This refined matching improves path planning accuracy by 20%-30%, avoiding mid-route stalls or detours caused by mismatches between vehicle performance and the environment, significantly enhancing task completion in complex scenarios.

[0058] 2. Dynamically avoid safety risks and reduce accident rates in extreme weather: This invention establishes a weather-vehicle risk matrix, converting meteorological data into specific constraints for safe vehicle driving, reducing the accident rate in extreme weather by 15% to 25%. At the same time, it reduces secondary delays caused by vehicle failures, building a solid safety line for emergency repair tasks.

[0059] 3. Optimize resource allocation efficiency and reduce scheduling costs: This invention uses a multi-objective optimization algorithm to incorporate vehicle load limits, battery range, maintenance equipment loading requirements, and real-time meteorological influencing factors into the model to achieve precise matching of tasks and vehicles. It also dynamically adjusts driving speed based on increased energy consumption caused by severe weather to avoid delays in refueling during the journey. It is estimated that the scheduling efficiency of equipment and manpower will be improved by 25% to 35%. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 is a flow chart of the method of the present invention;

[0062] Figure 2 This is a data architecture diagram of the present invention. DETAILED DESCRIPTION

[0063] The following describes in detail embodiments of the present application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application and are not to be construed as limiting the present application.

[0064] The present invention will be described in further detail below with reference to the accompanying drawings.

[0065] Combine Figure 1~Figure 2 The present invention provides a method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters, comprising:

[0066] S1. Build a multi-dimensional database. For meteorological data, integrate real-time weather APIs with historical weather data to extract macro-weather indicators such as precipitation and temperature, as well as micro-weather indicators such as road slipperiness, visibility, and wind speed. For vehicle data, establish a vehicle performance database, including drive mode, tire type, load capacity, and range.

[0067] S2. Improved algorithms and constraints. Dynamic weight allocation introduces path factors into route planning. Multi-objective optimization incorporates total travel time, weather risk, and vehicle loss into the objective function. Improved genetic algorithms balance the two objectives of "minimizing travel time" and "minimizing vehicle loss in extreme weather conditions" to improve route planning accuracy. Models are trained using historical data to predict optimal routes for different weather-vehicle combinations.

[0068] S3, dynamic adjustment and real-time feedback, deploy lightweight models on vehicle terminals to receive weather data in real time and adjust routes.

[0069] S4, Human-machine collaborative intervention: allows the driver to manually intervene in the path according to actual road conditions.

[0070] Preferably, S1 constructs a multidimensional database, and the specific steps are as follows:

[0071] S101. Build a meteorological database. Through standardized APIs, we acquire real-time, minute-by-minute weather data, covering basic weather indicators based on latitude and longitude, calculating micro-indicators, and analyzing weather warning information. We collect historical data from public meteorological databases, including hourly extreme weather events, corresponding road congestion durations, historical accident rates, and vehicle dispatch data from the vehicle management platform.

[0072] Data was cleaned and standardized to remove outliers, unify the time format, and convert spatial coordinates to WGS84 latitude and longitude. A tiered storage architecture was designed, using a time series database (InfluxDB) to store real-time data streams, supporting second-level queries. Macro and micro indicators were extracted using ETL tools and stored in a relational database. A spatiotemporal index (GIST index) was established to accelerate regional weather queries.

[0073] S103: Constructing a vehicle parameter database. Basic attributes such as vehicle drive mode, tire type, load, and range are obtained from the vehicle management platform's resource management module and encoded into a standardized dictionary. Real-time load data and other data are obtained from the vehicle management platform's dispatch data. This data is combined with information such as remaining battery life, tire pressure, and braking system information to construct a dynamic vehicle parameter profile. The average values ​​of vehicle performance parameters under various weather conditions are obtained at the test site to form a vehicle environmental adaptability matrix. Finally, a weather-vehicle parameter mapping table, a range attenuation model, and a load constraint library are established.

[0074] S104. Using "route ID + timestamp + vehicle ID" as the primary key, a real-time association table of weather data and vehicle parameters is established. A standardized calling interface is provided to support real-time acquisition of route factors during route optimization algorithms.

[0075] Preferably, S2 improves the algorithm and constraints, and the specific steps are as follows:

[0076] S201. Dynamic weight allocation: Introduce path factors in path planning. Sorting out the bad weather that affects vehicle travel, divided into 15 types, namely rainstorm, thunderstorm, heavy rain, typhoon, hurricane, tornado, strong wind, high temperature, hail, freezing, freezing, blizzard, heavy snow, haze, sandstorm and other weather. Path factor J i = path distance × ;W fa is the weather macro-index coefficient (rainstorm, thunderstorm, etc.), with a value of 0.3~0.6; W fb V is the weather micro-index coefficient (road slipperiness, visibility, etc.), with a value of 0.3~0.6; i U is the vehicle parameter coefficient (tire slip resistance coefficient, driving mode, etc.), assigned a value of 0.3~0.7; t The path coefficient defaults to 1. If the path is risky, the path coefficient increases. For example, in moderate rain or above, snow, ice, or freezing weather, and when the path slope is greater than or equal to 15°, the path coefficient is 1.5.

[0077] S202, Multi-Objective Optimization: Incorporate total travel time, weather risk, and vehicle loss into the objective function. By improving the genetic algorithm, balance the two objectives of "minimizing travel time" and "minimizing vehicle loss in extreme weather conditions" to improve route planning accuracy.

[0078] The formula of the initial scheduling model for path optimization is as follows:

[0079] ;

[0080] In the formula, X=(x1,x2,...,x m ), x1,x2,...,x m Where m is the vehicle loss dimension, f1(X) is the variable with the minimum total repair time, and f2(X) is the variable with the minimum vehicle loss under extreme weather conditions.

[0081] Furthermore, the calculation formula for the minimum variable f1(X) of the total path time is as follows:

[0082] ;

[0083] In the formula, P is the dispatch task, D is the corresponding route selection in each dispatch task, R is the path selection on the road, and J is the route selection on the road. i is the path factor, T i is the average time spent on the i-th path under normal weather conditions, J wk is the path waiting time, X vg As the constraint condition for repairing vehicles, each dispatch task is assigned at least one vehicle. Under the condition of no human intervention and normal road conditions, X vg=1.

[0084] Furthermore, the calculation formula for the minimum vehicle loss variable f2(X) under extreme weather conditions is as follows:

[0085] ;

[0086] In the formula, J i is the path factor, U i is the vehicle loss (tires, engines, braking systems, lights, etc.) caused by the adverse weather factors under the i-th path, L i is the loss accumulation factor of the i-th path (1 / 1.2).

[0087] Furthermore, the path optimization initial scheduling model in S202 has two pre-constraints:

[0088] Normal driving capability of emergency vehicles under extreme weather conditions Y i is an integer greater than zero, Y i This is the vehicle inspection data in the vehicle management platform. This is constraint condition 1, and the formula is as follows:

[0089] Y i ≥0;

[0090] Vehicle loss caused by extreme weather i Less than the delay loss H of the vehicle dispatch order i For constraint 2, the formula is as follows:

[0091] ;

[0092]

[0093]

[0094] Where α is the impact ratio of extreme weather.

[0095] On the premise that the above two constraints are met, the calculation of the path optimization initial scheduling model in S202 is performed.

[0096] Furthermore, the improved genetic algorithm is used to optimize the path optimization parameter variables, including:

[0097] A two-layer encoding scheme is used to encode the path optimization parameters in the genetic algorithm. The first layer uses integer encoding for the path factors, while the second layer converts the path factors into binary encoding. Dynamic weight adjustment through genetic operations enables adaptive allocation of path factors. Mutation of each binary bit fine-tunes the parameters, while crossover combines the advantages of different parameter combinations. Initial parameters related to path optimization are set, including the initial path, iteration number, and search count. The path optimization fitness function is determined to be the inverse of the objective function to ensure maximum fitness. A path selection strategy and terminating criteria are determined, followed by crossover and mutation operations. When the set number of iterations is reached, the target parameter variables are determined.

[0098] Furthermore, the selection strategy and stopping criteria are expressed as follows:

[0099] ;

[0100] ;

[0101] In the formula, W(x i ) is the probability of each individual being inherited to the next generation, U(x i ) is the probability of chromosome being selected.

[0102] S203. Based on the massive amount of historical repair task data in the vehicle management platform, including weather data and vehicle parameter data, data governance is performed according to the time dimension. The data is cleaned and normalized, key feature information is extracted, and a high-dimensional feature vector is constructed as the model input. A deep network architecture is adopted, with the path factor as the input layer and the path decision as the output layer. The state space is defined to include the current location, destination, real-time weather, and vehicle parameters; the behavior space is the optional path branch. The reward function is designed as the path optimization initial scheduling model. The pre-processed historical data is divided into a training set E = (e1, e2, ..., e m )、Validation set V=(v1,v2,...,v m ), and the test set T=(t1,t2,...,t m During training, the agent selects a path based on its current state in a simulated repair scenario, obtains reward feedback through interaction with the environment, continuously optimizes its strategy, monitors model performance using a validation set V, and adjusts hyperparameters to prevent overfitting until the model converges. The test set T is then used to evaluate the agent's ability to predict the optimal path.

[0103] Preferably, when S3 dynamically adjusts and provides real-time feedback, based on edge computing, during emergency repair tasks, the on-board terminal collects weather and vehicle data in real time, inputs the trained path optimization scheduling model, supports rapid output of the optimal path, and the driver drives according to the optimal path, and continuously feeds back data to dynamically optimize the model.

[0104] In the S4 human-machine collaborative mode, the driver is allowed to manually adjust the route based on actual road conditions. If the vehicle deviates from the optimal planned route by more than 500 meters or stops for more than 3 minutes, the system will automatically prompt "Do you want to adjust the route due to road conditions?" The driver confirms and initiates re-planning.

[0105] Preferably, in the S4 human-machine collaborative mode, the driver intervention instruction is converted into a dynamic emergency vehicle constraint condition X vg , integrated with real-time meteorological data and vehicle parameters to form a new state space. Using an improved genetic algorithm, the current vehicle position is used as a new starting point to search for feasible paths within the local road network, while retaining the valid sections of the original plan that were not affected, keeping the calculation time within 5 seconds. A method for optimizing emergency vehicle routes that integrates meteorological data and vehicle parameters dynamically adjusts the weight of the path factor J to prioritize the driver's intervention intentions while balancing the two goals of "shortest time" and "minimizing vehicle damage in extreme weather," generating alternative routes for the driver's secondary confirmation.

[0106] In order to more clearly illustrate the specific embodiment of the present invention, an embodiment is provided below:

[0107] A method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters, the method comprising the following steps:

[0108] S1. Construct a multi-dimensional database. In terms of meteorological data, integrate real-time meteorological API and historical weather data, extract macro weather indicators such as precipitation (precipitation_type) and temperature (temperature), and micro weather indicators such as road wetness (road_wetness), visibility (visibility), wind speed (gust_speed), and crosswind index (CrosswindIndex). In terms of vehicle data, establish a vehicle performance database, including drive mode (drive_type), tire type (tire_type_id), load capacity (load_capacity), endurance and energy consumption table (range_and_consumption), etc. A method for optimizing the path of emergency repair vehicles that integrates meteorological data and vehicle parameters, and its vehicle performance multi-dimensional database has one-to-one and one-to-many relationships. Taking tires as an example, there are multiple types of tires, and a new intermediate table vehicle_tire_relation (vehicle_id+tire_type_id) is added to handle many-to-many relationships. The data architecture diagram of the present invention is as follows: Figure 2 shown.

[0109] Preferably, during the meteorological database construction step, a standardized API interface is used to obtain real-time minute-by-minute meteorological data, covering basic weather indicators (precipitation intensity, temperature, humidity, wind speed, wind direction, air pressure, etc.) based on latitude and longitude, while also analyzing meteorological warning information. Historical data is collected from public meteorological databases, including hourly extreme weather events, corresponding road congestion duration, historical accident rates, and vehicle dispatch data from the vehicle management platform. Micro-meteorological indicators, including road slipperiness, visibility, and crosswind index, are calculated.

[0110] Optimally, during the data cleaning and standardization steps, outliers are removed, the time format is standardized, and spatial coordinates are converted to WGS84 latitude and longitude. A tiered storage architecture is designed, using a time series database (InfluxDB) to store real-time data streams and support second-level queries. Macro and micro indicators are extracted using ETL tools and stored in a relational database. A spatiotemporal index (GIST index) is established to accelerate regional weather queries.

[0111] Preferably, during the vehicle parameter database construction step, basic attributes such as vehicle drive mode, tire type, load, and range are obtained from the vehicle management platform's resource management module and encoded into a standardized dictionary. Real-time load and other data are obtained from the vehicle management platform's dispatch data and combined with information such as remaining battery life, tire pressure, and braking system to construct a dynamic vehicle parameter profile. The mean values ​​of vehicle performance parameters under various weather conditions are obtained at the test site to form a vehicle environmental adaptability matrix. Finally, a weather-vehicle parameter mapping table, a range attenuation model, and a load constraint library are established.

[0112] A real-time association table between weather data and vehicle parameters is established using "route ID + timestamp + vehicle ID" as the primary key. A standardized calling interface is provided to support real-time acquisition of route factors during route optimization algorithms.

[0113] S2. Improved algorithms and constraints. Dynamic weight allocation introduces path factors into route planning. Multi-objective optimization incorporates total travel time, weather risk, and vehicle loss into the objective function. Improved genetic algorithms balance the two objectives of "minimizing travel time" and "minimizing vehicle loss in extreme weather conditions" to improve route planning accuracy. Reinforcement learning utilizes historical data to train models and predict optimal routes for different weather and vehicle combinations.

[0114] Preferably, path factors are introduced in path planning. Severe weather conditions that affect vehicle travel are sorted out and divided into 15 types, namely, rainstorm, thunderstorm, heavy rain, typhoon, hurricane, tornado, strong wind, high temperature, hail, freezing, freezing, blizzard, heavy snow, haze, sandstorm and other weather conditions. In the calculation of path factors, W faThe coefficient of weather macro-index is 0.30 for thunderstorm, heavy rainfall, strong wind, high temperature, and haze (PM2.5≤250); 0.35 for freezing, condensation, heavy snow, rainstorm, sandstorm, and haze (PM2.5≥250); 0.4 for blizzard and hail; and 0.6 for typhoon, hurricane, and tornado. fb It is the weather micro-index coefficient, road slipperiness (assigned a value of 0.3-0.6 according to the classification), visibility (assigned a value of 0.3-0.6 according to the classification), and crosswind risk index (assigned a value of 0.3-0.6 according to the classification). i It is the vehicle parameter coefficient, including tire skid resistance coefficient (normal tire 0.5 / snow tire 0.3), drive mode (two-wheel drive 0.7 / four-wheel drive 0.3), load (≤5 tons 0.4 / ≥5 tons 0.7), cruising range (≤path*1.250.7 / ≥path*1.250.3), etc. t The default value is 1. If the path is risky, the path coefficient is increased. For example, in moderate rain or above, snow, ice, or freezing weather, with a path slope of 15° or greater, the path coefficient is 1.5. For heavy vehicles with a load exceeding 20 tons, the path coefficient is adjusted to 1.8 for high-risk paths such as bridges when wind speeds exceed level 6.

[0115] The total driving time f1(X), weather risk, and vehicle loss f2(X) are incorporated into the objective function. By improving the genetic algorithm, the two goals of "shortest time" and "minimizing vehicle loss in extreme weather" are balanced to improve the accuracy of route planning. The formula of the initial scheduling model for route optimization is:

[0116] ;

[0117] Where, X=(x1,x2,...,x m ), x1,x2,...,x m Where m is the vehicle loss dimension, f1(X) is the variable with the minimum total repair time, and f2(X) is the variable with the minimum vehicle loss under extreme weather conditions.

[0118] Furthermore, the path optimization initial scheduling model in S202 has two pre-constraints:

[0119] Normal driving capability of emergency vehicles under extreme weather conditions Y i is an integer greater than zero, Y i This is the vehicle inspection data in the vehicle management platform. This is constraint condition 1, and the formula is as follows:

[0120] Y i ≥0;

[0121] Vehicle loss caused by extreme weather i Less than the delay loss H of the vehicle dispatch orderi For constraint 2, the formula is as follows:

[0122] ;

[0123] ;

[0124] ;

[0125] Where α is the impact ratio of extreme weather.

[0126] On the premise that the above two constraints are met, the calculation of the path optimization initial scheduling model in S202 is performed.

[0127] Furthermore, an improved genetic algorithm is used to optimize the path optimization parameter variables. A two-layer encoding scheme is used to encode the path optimization parameters in the genetic algorithm: the first path layer uses integer encoding, and the second path factor is converted to binary encoding. Dynamic weight adjustment is achieved through genetic operations to achieve adaptive allocation of path factors. Mutating each binary bit fine-tunes the parameters, while crossover combines the advantages of different parameter combinations. Initial parameters related to path optimization are set, including the initial path, iteration number, and search count. The path optimization fitness function is determined to be the inverse of the objective function to ensure maximum fitness. A path selection strategy and terminating criteria are determined, followed by crossover and mutation operations. When the set number of iterations is reached, the target parameter variables are determined.

[0128] Preferably, based on the massive historical repair task data in the vehicle management platform, a high-dimensional feature vector X is constructed. t as model input. Where t represents the time dimension, D is the total feature dimension, X t =[X 天气宏观,t ;X 天气微观,t ;X 车辆参数 ;X 路径系数 ;X 时间维度 ]. 天气宏观,t =[d 事件,t ,s 强度,t ,T 温度 ,......];X 天气微观,t =[R 道路湿滑度 ,V 能见度 ,C 横风指数 ,δ 路面附着度 ,......];X 车辆参数 =[E drive mode, W 载重 ,L 续航 ,T 轮胎磨损 ,B 电池状态,......], etc. A deep network architecture is used, with path factors as the input layer and path decisions as the output layer. The state space is defined to include the current location, destination, real-time weather, and vehicle parameters; the behavior space is defined as the optional path branches. The reward function is designed as the path optimization initial scheduling model. The preprocessed historical data is divided into a training set E = (e1, e2, ..., e m )、Validation set V=(v1,v2,...,v m ), and the test set T=(t1,t2,...,t m During training, the agent selects a path based on its current state in a simulated repair scenario, obtains reward feedback through interaction with the environment, continuously optimizes its strategy, monitors model performance using a validation set V, and adjusts hyperparameters to prevent overfitting until the model converges. The test set T is then used to evaluate the agent's ability to predict the optimal path.

[0129] Preferably, during the dynamic adjustment and real-time feedback in step S3, differentiated parameters are set for two-wheel drive (2WD) and four-wheel drive (4WD) vehicles in heavy snow. In a certain road network example, after optimization, the total travel time for a two-wheel drive vehicle (optimal path A→C→E) is 81 minutes, vehicle loss is 72 loss units, and the dispatch delay loss is 90 loss units, resulting in a total target value of 243, a 15.1% reduction compared to the pre-optimization period. The total travel time for a four-wheel drive vehicle (A→C→D→E) is 62 minutes, vehicle loss is 57.2 loss units, and the dispatch delay loss is 78 risk units, resulting in a total target value of 197.6, a 8.2% reduction compared to the pre-optimization period.

[0130] Preferably, in step S4, in the human-machine collaborative mode, the driver is allowed to manually intervene in the route according to the actual road conditions. If the vehicle deviates from the optimal planned route by more than 500 meters or stops for more than 3 minutes, the system automatically prompts "Do you want to adjust the route due to road conditions?" and the driver confirms and initiates re-planning.

[0131] Preferably, in step S4, in the human-machine collaborative mode, the driver intervention instruction is converted into a dynamic emergency vehicle constraint condition X vg , (e.g. the weight of prohibited road sections is set to ∞, and the weight of driver intervention sections is reduced by 40%), and is integrated with real-time meteorological data (crosswind risk index is related to bridge path) and vehicle parameters (load change is related to path slope) to form a new state space. An improved genetic algorithm is used to search for feasible paths within the local road network with the current vehicle position as the new starting point, while retaining the valid sections that have not been intervened in the original plan, and the calculation time is controlled within 5 seconds. A repair vehicle path optimization method integrating meteorological data and vehicle parameters is based on the path factor J. i The path coefficient U involved in t, the weights can also be adjusted dynamically to give priority to the driver's intervention intention, while balancing the two goals of "shortest time" and "minimum vehicle loss in extreme weather", and generating alternative routes for the driver to confirm again.

[0132] Finally, all the parts not described in the present invention adopt mature products and mature technical means in the existing technology.

[0133] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters, characterized in that: The following steps are involved: S1. Build a multi-dimensional database: Integrate real-time weather data with historical weather data to extract macro-weather indicators and micro-road environment indicators; Build a vehicle performance database including drive mode, tire type, load capacity and range; S2. Improved path optimization algorithm, including: Dynamic weight allocation: Calculates the path factor by integrating weather coefficients including macro and micro weather indicators and vehicle performance parameters; Multi-objective optimization: Construct an objective function with total travel time, weather risk, and vehicle loss as variables, and solve the multi-objective problem using an improved genetic algorithm; S3, Dynamic route adjustment: Deploy a lightweight model on the vehicle terminal to receive real-time weather and vehicle data and update the optimal route; S4, Human-machine collaborative intervention: When the vehicle deviates from the planned path or stays for more than a threshold, path replanning is initiated, and an alternative path is generated in combination with the driver's manual adjustments; The algorithm and constraints in S2 need to be improved. The specific steps are as follows: S201. Dynamic Weight Allocation: Introducing path factors into route planning; sorting out severe weather conditions that affect vehicle travel and classifying them into 15 types: rainstorm, thunderstorm, heavy rainfall, typhoon, hurricane, tornado, high wind, high temperature, hail, freezing, freezing, blizzard, heavy snow, smog, and sandstorm; Where W fa is the weather macro-index coefficient, with a value of 0.3 to 0.6; W fb is the weather micro-index coefficient, with a value of 0.3 to 0.6; V i is the vehicle parameter coefficient, assigned a value of 0.3 to 0.7; U t is the path coefficient, which defaults to 1. When the corresponding path of the vehicle is risky, the path coefficient will be increased; S202, Multi-Objective Optimization: Incorporate total driving time, weather risk, and vehicle loss into the objective function; through an improved genetic algorithm, balance the two objectives of "shortest time" and "minimizing vehicle loss in extreme weather" to improve route planning accuracy; The formula of the initial scheduling model for path optimization is as follows: Where, X=(x1,x2,...,x m ), x1, x2, ..., x m , m is the vehicle loss dimension, f1(X) is the minimum variable for total repair time, and f2(X) is the minimum variable for vehicle loss under extreme weather conditions; S203. Based on the massive amount of historical repair task data in the vehicle management platform, including weather data and vehicle parameter data, data governance is performed according to the time dimension. The data is cleaned and normalized, key feature information is extracted, and a high-dimensional feature vector is constructed as the model input. A deep network architecture is adopted, with the path factor as the input layer and the path decision as the output layer. The state space is defined to include the current location, destination, real-time weather, and vehicle parameters; the behavior space is the optional path branch; the reward function is designed as the path optimization initial scheduling model; the preprocessed historical data is divided into a training set E = (e1, e2, ..., e m ), validation set V = (v1, v2, ..., v m ), and the test set T = (t1, t2, ..., t m During training, the agent selects a path based on its current state in a simulated repair scenario, obtains reward feedback through interaction with the environment, continuously optimizes its strategy, monitors model performance using a validation set V, and adjusts hyperparameters to prevent overfitting until the model converges. The agent then uses a test set T to evaluate its ability to predict the optimal path. The calculation formula of the minimum variable f1(X) of the total path time in S202 is as follows: Where P is the dispatch task, D is the corresponding route selection in each dispatch task, R is the path selection on the road, and J is the route selection on the road. i is the path factor, T i is the average time spent on the i-th path under normal weather conditions, J wk is the path waiting time, X vg As the constraint condition for repairing vehicles, each dispatch task is assigned at least one vehicle. Under the condition of no human intervention and normal road conditions, X vg =1; The calculation formula for the minimum vehicle loss variable f2(X) under the influence of extreme weather in S202 is as follows: Where, J i is the path factor, U i is the vehicle loss caused by the adverse weather factors on the i-th path, L i is the loss accumulation factor of the i-th path; The path optimization initial scheduling model in S202 has two pre-constraints: Normal driving capability of emergency vehicles under extreme weather conditions Y i is an integer greater than zero, Y i This is the vehicle inspection data in the vehicle management platform. This is constraint condition 1. The formula is as follows: Y i ≥0； Vehicle loss caused by extreme weather i Less than the delay loss H of the vehicle dispatch order i For constraint 2, the formula is as follows: a*h i <H i ; h i =∑(min(unit price of component*0.6, repair cost)*probability of damage*number of damaged components); H i = Number of repair team members * average hourly wage * duration of delay + duration of delay * probability * potential failure loss per unit; Where α is the impact ratio of extreme weather; On the premise that the above two constraints are met, the calculation of the path optimization initial scheduling model in S202 is performed.

2. The method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters according to claim 1, characterized in that: The specific steps of constructing the multidimensional database in S1 are as follows: S101. Build a meteorological database. Through standardized APIs, obtain minute-by-minute meteorological data in real time, covering basic weather indicators based on latitude and longitude, calculate micro-weather indicators, analyze meteorological warning information, and collect historical data from public meteorological databases, including hourly extreme weather events, corresponding road congestion duration, historical accident rates, and vehicle dispatch data from the vehicle management platform. S102, data cleaning and standardization, removing outliers, unifying time formats, and converting spatial coordinates to WGS84 longitude and latitude; Design a hierarchical storage architecture, using a time-series database to store real-time data streams and support second-level queries. Use ETL tools to extract macro and micro indicators, store them in a relational database, and establish spatiotemporal indexes to accelerate regional weather queries. S103: Constructing a vehicle parameter database. Basic attributes such as vehicle drive mode, tire type, load, and range are obtained from the vehicle management platform's resource management module and encoded into a standardized dictionary. Real-time load and other data are obtained from the vehicle management platform's dispatch order data. This data is combined with remaining battery life, tire pressure, and brake system information to construct a dynamic vehicle parameter profile. Average vehicle performance parameters under various weather conditions are obtained through test sites to form a vehicle environmental adaptability matrix. Finally, a weather-vehicle parameter mapping table, a range attenuation model, and a load constraint library are established. S104. Using "route ID + timestamp + vehicle ID" as the primary key, establish a real-time association table between weather data and vehicle parameters; provide a standardized calling interface to support real-time acquisition of route factors during route optimization algorithms.

3. The method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters according to claim 1, characterized in that: In S202, the improved genetic algorithm is used to optimize the path optimization parameter variables, and the optimization content includes: A two-layer coding mode is used to encode the path optimization parameters in the genetic algorithm. The first layer of the path layer uses integer coding, and the second layer of the weather-vehicle coupling factor is converted into binary coding. Dynamically adjust weights through genetic operations to achieve adaptive allocation of path factors; The mutation of each binary bit can fine-tune the parameters, and crossover can combine the advantages of different parameter combinations; Set the initial parameters related to path optimization, including the initial path, iterations, and search times; Determine the path optimization fitness function as the inverse of the objective function to ensure maximum fitness; Determine the strategy for selecting the path and the criteria for stopping iteration, then perform crossover and mutation operations. When the set number of iterations is reached, determine the target parameter variable.

4. The method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters according to claim 1, characterized in that: The selection strategy and stopping criteria in S202 are expressed as follows: Where, W(x i ) is the probability of each individual being inherited to the next generation, U(x i ) is the probability of chromosome being selected.

5. The method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters according to claim 1, characterized in that: In the aforementioned S3, edge computing is required. During emergency repair tasks, the on-board terminal collects weather and vehicle data in real time and transmits it to the trained path optimization scheduling model to support the rapid output of the optimal path. The driver drives according to the optimal path and continuously feeds back data to dynamically optimize the model.

6. The method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters according to claim 1, characterized in that: In the S4, in the human-machine collaborative mode, the driver is allowed to manually intervene in the route according to the actual road conditions; when the vehicle deviates from the optimal planned path by more than 500 meters or stops for more than 3 minutes, the system automatically prompts "Do you want to adjust the route due to road conditions?" and the driver confirms and starts re-planning.

7. The method for optimizing emergency vehicle routes by integrating meteorological data and vehicle parameters according to claim 6, characterized in that: In the S4, in the human-machine collaborative mode, the driver intervention instruction is converted into a dynamic emergency vehicle constraint condition X vg , and integrate with real-time meteorological data and vehicle parameters to form a new state space; using an improved genetic algorithm, taking the current vehicle position as a new starting point, searching for feasible paths within the local road network, while retaining the effective sections that have not been interfered with in the original plan, the calculation time is controlled within 5 seconds; based on the path factor J i The path coefficient U involved in t , dynamically adjust the weights, give priority to meeting the driver's intervention intention, and balance the two goals of "shortest time" and "minimum vehicle loss in extreme weather", and then generate alternative paths for the driver to confirm again.

Images