Internet of Things-based transportation route optimization methods and systems

Abstract

This application discloses a transportation route optimization method and system based on the Internet of Things (IoT), relating to the field of IoT technology. The method includes: loading real-time multi-source sensing data of the transportation environment; constructing a multi-objective dynamic optimization function; extracting environmental features from the real-time multi-source sensing data; performing adaptive adjustment of multi-objective weights to obtain the dynamic weight coefficients at the current moment; performing Pareto optimal path solving to obtain a candidate path set; loading a constraint set; detecting the conflict relationship between each candidate path and the constraint set; when there is a conflict between at least one of quasi-rigid constraints and elastic constraints, performing constraint relaxation and compensation cost calculation to obtain the relaxed path and the corresponding compensation cost; and selecting the optimal path scheme from the candidate path set or the relaxed path and outputting it. This solves the technical problems in existing technologies where fixed multi-objective weights cannot adapt to environmental changes and the constraint handling is too simple, leading to excessive shrinkage of the feasible solution space.

Description

Technical Field

[0001] This application relates to the field of Internet of Things (IoT) technology, specifically to IoT-based transportation route optimization methods and systems. Background Technology

[0002] With the development of IoT technology, transportation route optimization has evolved from static planning to dynamic real-time planning. Traditional transportation route planning methods mainly rely on static historical data and preset rules, which are difficult to adapt to the complex and ever-changing actual transportation environment. In actual transportation, road conditions change in real time, vehicle status fluctuates dynamically, and customer needs are constantly adjusted, making it difficult for fixed route planning schemes to achieve optimal results.

[0003] In existing technologies, the weights of each objective in the multi-objective optimization of transportation routes are fixed or determined by human experience, and cannot be dynamically adjusted according to real-time environmental characteristics. This leads to an excessive pursuit of timeliness in congested sections and a neglect of cost savings during less congested periods, resulting in poor overall optimization performance. Furthermore, the handling of constraints is relatively simple, usually treating all types of constraints equally. Once a candidate path conflicts with the constraints, it is directly discarded, failing to distinguish between rigid and adjustable constraints. This causes an excessive contraction of the feasible solution space and may miss better alternatives. Summary of the Invention

[0004] This application provides a transportation route optimization method and system based on the Internet of Things, which solves the technical problems in the prior art where fixed multi-objective weights cannot adapt to environmental changes and simple constraint handling leads to excessive shrinkage of the feasible solution space.

[0005] The technical solution to the above-mentioned technical problems in this application is as follows:

[0006] In a first aspect, this application provides a transportation route optimization method based on the Internet of Things, the method comprising:

[0007] Load real-time multi-source perception data of the transportation environment to obtain real-time multi-source perception data, wherein the real-time multi-source perception data includes road condition dynamic data, vehicle status data and task constraint data;

[0008] Based on the real-time multi-source sensing data, a multi-objective dynamic optimization function is constructed to obtain the multi-objective dynamic optimization function, wherein the multi-objective dynamic optimization function includes cost target, timeliness target and carbon emission target;

[0009] Environmental features are extracted from the real-time multi-source sensing data, and adaptive adjustment of multi-objective weights is performed based on the environmental features to obtain the dynamic weight coefficients at the current moment.

[0010] Based on the dynamic weighting coefficients and the multi-objective dynamic optimization function, Pareto optimal path solving is performed to obtain a candidate path set;

[0011] Load the constraint set of the transportation environment to obtain the constraint set, wherein the constraint set includes rigid constraints, quasi-rigid constraints and elastic constraints;

[0012] The conflict relationship between each candidate path in the candidate path set and the constraint condition set is detected to obtain the conflict detection result; when there is at least one conflict between quasi-rigid constraints and elastic constraints in the conflict detection result, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost.

[0013] Based on the dynamic weight coefficients, the conflict detection results, and the compensation cost, the optimal path scheme is selected and output from the candidate path set or the relaxed path.

[0014] Secondly, this application provides an Internet of Things-based transportation route optimization system, including:

[0015] The data acquisition module is used to load real-time multi-source sensing data of the transportation environment and obtain real-time multi-source sensing data, wherein the real-time multi-source sensing data includes road condition dynamic data, vehicle status data and task constraint data;

[0016] The function construction module is used to construct a multi-objective dynamic optimization function based on the real-time multi-source sensing data, and obtain the multi-objective dynamic optimization function, wherein the multi-objective dynamic optimization function includes cost target, timeliness target and carbon emission target;

[0017] The feature extraction module is used to extract environmental features from the real-time multi-source sensing data, perform adaptive adjustment of multi-objective weights based on the environmental features, and obtain the dynamic weight coefficients at the current moment.

[0018] The path optimization module is used to perform Pareto optimal path solving based on the dynamic weight coefficients and the multi-objective dynamic optimization function to obtain a set of candidate paths.

[0019] The constraint condition acquisition module is used to load the constraint condition set of the transportation environment and obtain the constraint condition set, wherein the constraint condition set includes rigid constraints, quasi-rigid constraints and elastic constraints;

[0020] The path detection module is used to detect the conflict relationship between each candidate path in the candidate path set and the constraint condition set, and obtain the conflict detection result; when there is at least one conflict between quasi-rigid constraints and elastic constraints in the conflict detection result, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost.

[0021] The scheme acquisition module is used to select the optimal path scheme from the candidate path set or the relaxed path based on the dynamic weight coefficient, the conflict detection result and the compensation cost.

[0022] This application provides one or more technical solutions, which have at least the following technical effects or advantages:

[0023] This application provides an IoT-based transportation route optimization method and system. First, by dynamically loading and processing real-time multi-source sensing data, a comprehensive digital mapping of the transportation environment is achieved, enabling route planning to make decisions based on the latest road conditions, vehicle status, and task requirements. Second, an environmental feature-driven multi-objective weight adaptive adjustment mechanism is adopted, overcoming the limitations of traditional fixed weights or manually set weights. This allows the priority of objectives such as cost, timeliness, and carbon emissions to be intelligently adjusted according to environmental changes, improving the environmental adaptability of multi-objective optimization. Third, through a hierarchical processing strategy of rigid constraints, quasi-rigid constraints, and flexible constraints, the relaxation benefits and compensation costs of adjustable constraints are quantitatively evaluated under the premise of ensuring traffic safety and regulatory compliance, expanding the feasible solution space and avoiding the loss of optimization opportunities due to simply abandoning conflicting paths. Finally, by combining Pareto optimality solutions and constraint relaxation mechanisms, a balance between global optimality and local flexibility is achieved. Through a comprehensive evaluation of dynamic weight coefficients, conflict detection results, and compensation costs, the output optimal route solution is ensured to be both economical and feasible in complex and ever-changing real-world transportation environments.

[0024] Through the above technical solutions, this application achieves intelligent, dynamic and refined transportation route optimization based on the fusion processing of real-time multi-source sensing data, adaptive dynamic adjustment of multi-objective weights, refined management of hierarchical constraints, and quantitative evaluation of constraint relaxation and compensation costs, effectively improving the decision-making quality and execution effect in complex transportation environments. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart illustrating the transportation route optimization method based on the Internet of Things provided in this application embodiment;

[0027] Figure 2 This is a schematic diagram of the structure of the Internet of Things-based transportation route optimization system provided in the embodiments of this application.

[0028] The components represented by each number in the attached diagram are explained below:

[0029] Data acquisition module 11, function construction module 12, feature extraction module 13, path optimization module 14, constraint acquisition module 15, path detection module 16, and solution acquisition module 17. Detailed Implementation

[0030] This application provides a transportation route optimization method and system based on the Internet of Things, which addresses the technical problems in the prior art where fixed multi-objective weights cannot adapt to environmental changes and simple constraint handling leads to excessive shrinkage of the feasible solution space.

[0031] Example 1, as Figure 1 As shown, this application provides a transportation route optimization method based on the Internet of Things, including:

[0032] S10: Load real-time multi-source perception data of the transportation environment to obtain real-time multi-source perception data, wherein the real-time multi-source perception data includes road condition dynamic data, vehicle status data and task constraint data;

[0033] In this embodiment, road condition dynamic data is obtained through multi-source collection via IoT sensor networks deployed along the road, roadside units (RSUs), and third-party traffic information platforms; vehicle status data is obtained in real time via onboard OBD interface, CAN bus, and onboard intelligent terminal; and task constraint data comes from order information and customer customization requirements from the logistics management system.

[0034] The multi-source sensing data is aggregated at the edge via the MQTT or CoAP protocol. After data cleaning, time synchronization and format standardization, it is stored in a distributed time-series database with a unified data structure.

[0035] Specifically, step S10 in the method includes:

[0036] Obtain the dynamic traffic data, which includes real-time congestion index, average traffic speed, traffic accident information, and weather conditions;

[0037] Obtain the vehicle status data, wherein the vehicle status data includes vehicle location, remaining driving range, load status and driving time;

[0038] Obtain the task constraint data, which includes customer time window, order urgency, and regional environmental protection requirement level.

[0039] In this embodiment, firstly, roadside sensing devices deployed on highways, urban arterial roads, and logistics parks collect raw data such as traffic flow, average vehicle speed, and road occupancy at fixed time intervals. After preprocessing by edge computing nodes, a real-time congestion index is generated. At the same time, traffic accident information interfaces released by traffic management departments and weather data from meteorological service platforms are accessed to form a complete description of dynamic road conditions.

[0040] Secondly, the acquisition of vehicle status data relies on the vehicle-mounted Internet of Things terminal. The GPS / BeiDou dual-mode positioning module transmits vehicle location information in real time. Combined with the battery management system or fuel level sensor, the remaining driving range is calculated. The load sensor dynamically monitors changes in cargo weight and records engine running time as the basis for accumulating driving time.

[0041] Secondly, task constraint data is extracted from the enterprise resource planning system and transportation management system. The expected arrival time interval in the customer order is parsed to generate the customer time window. The urgency of the order is determined according to the priority label marked on the order, and the environmental protection requirement level of the destination area is obtained by matching the environmental protection restriction policy database.

[0042] S20: Construct a multi-objective dynamic optimization function based on the real-time multi-source sensing data to obtain the multi-objective dynamic optimization function, wherein the multi-objective dynamic optimization function includes cost target, timeliness target and carbon emission target;

[0043] In this embodiment, the cost target comprehensively calculates fuel consumption costs, toll fees, vehicle depreciation costs, and driver labor costs. Fuel consumption costs are dynamically estimated based on the congestion index and average traffic speed in real-time traffic data. The fuel consumption coefficient for congested road sections is increased by 15% to 30%, while the standard fuel consumption is used for uncongested road sections. The timeliness target uses customer time window satisfaction rate and average on-time delivery rate as core indicators, incorporating the waiting costs for deliveries earlier than the lower limit of the customer's expected time window and the delay penalty costs for deliveries later than the upper limit of the time window into a unified quantitative framework. The carbon emission target establishes an emission factor model based on vehicle type, load status, and real-time driving conditions, distinguishing the different emission characteristics of pure electric, hybrid, and fuel vehicles. The carbon emissions of fuel vehicles are calculated by multiplying the product of mileage and comprehensive fuel consumption by the standard carbon emission factor, while the indirect carbon emissions from electricity sources are taken into account for new energy vehicles.

[0044] Secondly, a multi-objective dynamic optimization function is constructed based on real-time multi-source sensing data. For example, the mathematical expression of the multi-objective dynamic optimization function is as follows: , where x represents the path decision variable, t represents the current time, and each component function corresponds to the cost, timeliness and carbon emission targets respectively. In order to deal with the dimensional differences and conflicting relationships among multiple objectives, a normalization process based on the ideal point method is adopted to map the values ​​of each objective function to the interval [0,1], which facilitates subsequent weighted aggregation and Pareto front solution.

[0045] S30: Extract environmental features from the real-time multi-source sensing data, perform adaptive adjustment of multi-objective weights based on the environmental features, and obtain the dynamic weight coefficients at the current moment;

[0046] In this embodiment, environmental features are extracted from real-time multi-source sensing data, reflecting a dynamic set of parameters reflecting the urgency of the transportation task and external traffic conditions. Based on these environmental features, adaptive adjustments to multi-objective weights are performed. This embodiment employs a hybrid decision-making mechanism combining fuzzy inference and reinforcement learning, processing environmental features to output the dynamic weight coefficients for the current moment.

[0047] Furthermore, environmental features are extracted from the real-time multi-source sensing data, and adaptive adjustment of multi-objective weights is performed based on the environmental features to obtain the dynamic weight coefficients at the current moment, including the following:

[0048] Collect historical environmental feature sequences and historical optimal weight sequences under preset transportation scenarios, wherein the historical environmental feature sequences include historical congestion index sequences, historical order urgency sequences, and historical weather level sequences;

[0049] Using the historical environmental feature sequence as input and the historical optimal weight sequence as supervision, a weight adaptive adjustment model is trained to obtain the weight adaptive adjustment model, wherein the weight adaptive adjustment model adopts a fuzzy neural network or a deep reinforcement learning network.

[0050] The weight adaptive adjustment model is associated with and stored with the preset transportation scenario, and added to the weight adjustment model library;

[0051] Based on the current transportation scenario, the adaptive weight adjustment model in the weight adjustment model library is invoked to process the environmental features in the real-time multi-source perception data and output the dynamic weight coefficient at the current moment.

[0052] In this embodiment, firstly, historical environmental feature sequences and historical optimal weight sequences are collected under preset transportation scenarios. The historical environmental feature sequences cover the changing patterns of road congestion indices in different seasons, time periods, and regions. The historical order urgency sequence reflects the distribution of task urgency under typical scenarios such as e-commerce promotions and holidays. The historical weather level sequence records the degree of impact of severe weather such as heavy rain, fog, and snow on transportation efficiency. The historical optimal weight sequence is obtained through offline simulation optimization. That is, under the condition of complete environmental information, existing enumeration algorithms are used to solve for the target weight combination that maximizes overall satisfaction in each historical scenario, forming a training sample set with supervised labels.

[0053] Secondly, using historical environmental feature sequences as input and historical optimal weight sequences as supervision, an adaptive weight adjustment model is trained. When using a fuzzy neural network, a five-layer network structure is designed: the input layer receives the normalized environmental feature vector; the fuzzification layer uses a Gaussian membership function to transform the input into linguistic variables; the rule layer constructs a fuzzy inference rule base between environmental features and weight adjustments; the defuzzification layer outputs continuous weight coefficients; and the output layer is normalized to ensure that the sum of the weights of each objective is 1. When using a deep reinforcement learning network, a Markov decision process is constructed with environmental features as the state space, weight adjustments as the action space, and multi-objective comprehensive optimization effects as immediate rewards. Existing proximal policy optimization algorithms are used to train the policy network, enabling the agent to learn to output the optimal weight adjustment strategy under different environmental states.

[0054] Markov decision-making refers to a mathematical framework in which an agent makes sequential decisions in an environment through state transitions, action selections, and reward feedback. Its core elements include state space S, action space A, state transition probability P, reward function R, and discount factor γ. In this embodiment, state space S is composed of environmental feature vectors such as normalized real-time congestion index, order urgency level, weather impact coefficient, remaining driving range ratio, and customer time window urgency. Action space A is defined as a three-dimensional continuous space, with each dimension corresponding to the weight adjustment of cost target, timeliness target, and carbon emission target, respectively. The adjustment range is [-0.2, 0.2] and satisfies the non-negative constraint of the adjusted weights. The reward function R is designed as an immediate evaluation of the comprehensive optimization effect. When the adjusted weight coefficients are used to solve the path and obtain the Pareto solution set, the weighted combination of the hypervolume index and congestion index of the solution set is used as a positive reward. At the same time, a constraint violation penalty term is introduced. If the generated path solution has a rigid constraint conflict, a negative reward is given.

[0055] The state transition probability P describes the evolution of the environmental state after weight adjustment. Since the dynamic changes of the transportation environment are random and non-stationary, this application adopts an empirical transition model based on historical data statistics. That is, the frequency distribution of historical state transitions under the same or similar environmental characteristics is used to approximate the true transition probability. For example, the discount factor γ is set to 0.95 to balance the importance of immediate rewards and long-term cumulative rewards, and to ensure that the agent achieves a reasonable trade-off between short-term optimization effect and long-term strategy stability.

[0056] Next, the trained weight adaptive adjustment model is associated with and stored with the preset transportation scenario identifiers and added to the weight adjustment model library. The preset transportation scenarios are classified in multiple dimensions according to regional characteristics, business type and time period attributes, such as cold chain distribution scenario in the Yangtze River Delta region, express delivery scenario in the Pearl River Delta region, and bulk freight scenario in the Beijing-Tianjin-Hebei region. Each scenario corresponds to an independent model instance to adapt to different decision preferences.

[0057] Furthermore, real-time extracted environmental features such as congestion index, order urgency, and weather level are input into the model for the corresponding scenario. The model outputs cost target weight w1(t), timeliness target weight w2(t), and carbon emission target weight w3(t), satisfying... Furthermore, all weights are non-negative. For example, in scenarios where order urgency surges during e-commerce promotions, the model automatically increases the weight of the timeliness target to above 0.5; in scenarios involving travel in areas with strict environmental regulations, the weight of the carbon emission target is adjusted accordingly; and in normal scenarios with unobstructed roads and ample time, the cost target takes the lead, thereby achieving environmental adaptation with multiple target priorities.

[0058] Furthermore, based on the environmental features in the real-time multi-source sensing data, adaptive adjustment of multi-objective weights is performed to obtain the dynamic weight coefficients at the current moment, followed by:

[0059] The dynamic weight coefficients at the current time and the previous time are smoothed by exponential moving average to obtain smoothed dynamic weight coefficients, and the smoothed dynamic weight coefficients are used as the updated dynamic weight coefficients at the current time.

[0060] In this embodiment, the dynamic weight coefficient at the current moment and the dynamic weight coefficient at the previous moment are subjected to exponential moving average smoothing, which can effectively suppress the drastic fluctuations in weight caused by environmental noise and avoid decision oscillations caused by frequent switching of optimization targets in a short period of time.

[0061] Specifically, the formula for calculating the exponential moving average smoothing is as follows: Where w(t) is the original weight coefficient output by the model at the current time, and w(t-1) is the smoothed weight coefficient at the previous time. The smoothing coefficient α ranges from 0.3 to 0.7. The larger the value of α, the more sensitive the response to changes in the current environment. The smaller the value of α, the smoother the weight adjustment. The value of α is dynamically selected according to the stability requirements of the transportation task. For example, for cold chain transportation tasks with extremely high time requirements, α=0.6 is taken, and for bulk freight tasks that are sensitive to costs, α=0.4 is taken, so as to achieve fine control between balancing response speed and decision stability.

[0062] S40: Based on the dynamic weight coefficients and the multi-objective dynamic optimization function, perform Pareto optimal path solving to obtain a candidate path set;

[0063] In this embodiment, Pareto optimal path solving is performed based on dynamic weight coefficients and multi-objective dynamic optimization functions, using an improved multi-objective evolutionary algorithm. This algorithm is an adaptive improvement on the traditional non-dominated sorting genetic algorithm NSGA-II, designed to address the discrete characteristics and dynamic constraints of the transportation path optimization problem.

[0064] First, a chromosome representation method based on real number encoding is adopted, where each chromosome corresponds to a complete transport path, and the gene positions are recorded sequentially with the sequential numbers of the nodes passed through. An initialization strategy combining random generation, nearest neighbor heuristic construction, and 2-opt local search is used to ensure the wide distribution of the population in the solution space and preliminary quality assurance.

[0065] Secondly, adaptive crossover and mutation operators are designed. The crossover operation uses the ordered crossover OX operator to maintain the legality of the path. The mutation operation integrates three modes: exchange mutation, insertion mutation, and reverse mutation. The mutation probability is dynamically adjusted according to the number of generations of population evolution. For example, in the early stage, the mutation probability is set to 0.15 to focus on global exploration, and in the later stage, the mutation probability is reduced to 0.05 to focus on local development.

[0066] Furthermore, a preference guidance mechanism based on dynamic weight coefficients is introduced. In the environment selection stage, not only is the Pareto dominance relationship considered, but also the cosine of the angle between each non-dominated solution and the ideal direction guided by the weight vector is calculated. Solutions with higher matching degree with the current dynamic weight coefficients are retained first, so that the search direction of the algorithm can be adaptively adjusted with changes in the environment.

[0067] Furthermore, to address the issue of hard constraint violations in transportation routes, a feasibility rule-based approach is adopted. The degree of constraint violation is quantified into a penalty function value. The fitness value of any infeasible solution is inferior to that of a feasible solution. At the same time, feasible solutions are sorted according to the objective function value to ensure that the final output set of candidate routes consists entirely of feasible solutions that satisfy rigid constraints such as vehicle capacity, time window, and driving range.

[0068] For example, by setting a maximum number of iterations of 200 generations or a convergence threshold of less than 0.1% for 20 consecutive generations of Pareto front hypervolume change rate as a termination condition, a set of candidate paths containing 50 to 100 non-dominated solutions is output, which forms a Pareto front with a balanced distribution in terms of cost, timeliness and carbon emissions.

[0069] S50: Load the constraint set of the transportation environment to obtain the constraint set, wherein the constraint set includes rigid constraints, quasi-rigid constraints and elastic constraints;

[0070] In this application embodiment, rigid constraints refer to inviolable physical limits and regulatory red lines, including the maximum load limit of the vehicle, the upper limit of the remaining driving range of the fuel tank or battery, the legal upper limit of 4 hours of continuous driving time for the driver, and the hard cutoff time of the customer's time window. Any path scheme that violates the rigid constraints is directly determined to be infeasible.

[0071] Quasi-rigid constraints refer to constraints that allow limited breaches under specific conditions but require additional costs or risks. For example, in emergency delivery scenarios, temporary passes can be applied for under urban environmental protection traffic restrictions, but expedited approval fees must be paid and the issue will be included in the company's credit record. Although early arrival within the customer's expected time window does not constitute a breach of contract, it will incur energy costs for refrigerated vehicles to pre-cool.

[0072] Elastic constraints refer to preference constraints that can be flexibly adjusted through economic levers. These include the willingness to compress the total mileage of the route, the preference for choosing between highways and national and provincial roads, and the preference for driving at night versus during the day. The degree to which elastic constraints are met directly affects the calculation of cost components in the objective function, but does not determine the feasibility of the solution.

[0073] Furthermore, the loading process of the constraint set converges from multiple sources, including the enterprise rule engine, the industry regulatory database, and the real-time policy interface. The enterprise rule engine stores internal vehicle dispatching procedures and customer contract terms, the industry regulatory database synchronizes with the latest regulations and standards issued by the Ministry of Transport and the Ministry of Ecology and Environment, and the real-time policy interface connects with temporary control notices from traffic police and road administration departments in various regions, forming a hierarchical and scalable constraint knowledge graph.

[0074] S60: Detect the conflict relationship between each candidate path in the candidate path set and the constraint condition set, and obtain the conflict detection result; when there is at least one conflict between quasi-rigid constraints and elastic constraints in the conflict detection result, perform constraint relaxation and compensation cost calculation to obtain the relaxed path and the corresponding compensation cost.

[0075] In this embodiment, the conflict relationship between each candidate path in the candidate path set and the constraint condition set is detected. A hierarchical and progressive constraint detection strategy is adopted. Conflict identification is performed in the order of priority of rigid constraints, quasi-rigid constraints and elastic constraints. If at least one of quasi-rigid constraints and elastic constraints exists in the conflict detection result, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost.

[0076] Specifically, detecting the conflict relationship between each candidate path in the candidate path set and the constraint set to obtain conflict detection results includes:

[0077] Load the rigid constraints to obtain a set of rigid constraints, wherein the rigid constraints include traffic rule speed limits, restricted areas, and bridge load-bearing limits;

[0078] When any candidate path violates any rigid constraint in the set of rigid constraints, the candidate path is marked as an infeasible path and removed from the set of candidate paths to obtain an updated set of candidate paths.

[0079] Load the quasi-rigid constraints to obtain a set of quasi-rigid constraints, wherein the quasi-rigid constraints include customer time windows and driving duration regulations;

[0080] Load the elastic constraints to obtain a set of elastic constraints, wherein the elastic constraints include route preferences and stop order;

[0081] The degree of deviation between each candidate path in the updated candidate path set and the quasi-rigid constraint set and the elastic constraint set is recorded as the conflict detection result.

[0082] In this embodiment, firstly, a set of rigid constraints is loaded for initial screening. Candidate paths that violate absolutely insurmountable conditions such as traffic speed limits, restricted areas, and bridge load-bearing limits are directly eliminated, ensuring that the remaining paths all meet the basic requirements of physical safety and regulatory compliance. For example, if a candidate path plans to pass through an underground passage with a height limit of 3.5 meters, but the actual vehicle height is 4.2 meters, the path is immediately marked as infeasible and removed from the set; if the path includes urban expressway sections where hazardous chemical vehicles are prohibited during restricted hours, the same forced elimination operation is performed.

[0083] Secondly, for the candidate paths that pass the rigid constraint test, a quasi-rigid constraint set and an elastic constraint set are further loaded, and the deviation time from the customer's expected time window, the extent of exceeding the driver's continuous driving time limit, and the degree of difference from the preset route preference are calculated for each path.

[0084] Specifically, the degree of deviation is quantified using a standardized scoring mechanism. Deviation within a time window is counted as one deviation unit for every hour earlier or later than the time window, and deviation exceeding the driving time is counted as one deviation unit for every 15 minutes exceeding the time window. Deviation from route preference is calculated based on the overlap ratio between the actual driving route and the preferred route, forming a multi-dimensional conflict detection matrix.

[0085] Furthermore, when at least one of the quasi-rigid constraints and elastic constraints exists in the conflict detection results, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost, including:

[0086] Extract the quasi-rigid or elastic constraints that have conflicted from the conflict detection results and set them as constraints to be relaxed.

[0087] Calculate the relaxation benefit of the constraint to be relaxed and obtain the relaxation benefit value. The calculation basis of the relaxation benefit includes time saving, energy consumption reduction and task completion rate improvement.

[0088] Calculate the compensation cost of the constraint to be relaxed and obtain the compensation cost value. The calculation basis of the compensation cost includes time window delay penalty, detour additional cost, and additional cost of change of transportation mode.

[0089] When the relaxation benefit value is greater than the compensation cost value, a relaxation operation is performed on the constraint to be relaxed, a relaxed path is generated, and the compensation cost value is associated with the relaxed path.

[0090] When the relaxation benefit value is less than or equal to the compensation cost value, the relaxation operation is abandoned and the original candidate path is maintained.

[0091] In this embodiment, firstly, the quasi-rigid or elastic constraints that have deviated from the conflict detection results are extracted as the objects to be evaluated. The quantitative ratio of the relaxation benefit to the compensation cost for each constraint to be relaxed is then calculated. The calculation of the relaxation benefit considers the time efficiency gains resulting from time savings, the cost optimization resulting from reduced energy consumption, and the improved customer satisfaction resulting from increased task completion rate. For example, extending the customer's expected time window from 9:00 AM to 11:00 AM to 8:00 AM to 12:00 PM can shorten the detour distance by 35 kilometers, corresponding to a fuel cost reduction of approximately 28 yuan, while simultaneously reducing carbon emissions by 8.5 kilograms. The overall relaxation benefit is assessed at 42 benefit units.

[0092] Secondly, the calculation of compensation costs covers the loss of customer satisfaction caused by the time window delay, the additional fuel and labor costs caused by detours, and the decrease in loading and unloading efficiency caused by changes in transportation methods. Taking the aforementioned time window relaxation as an example, the energy consumption cost of pre-cooling waiting caused by arriving early is 15 yuan, and the expected cost of the potential default risk of arriving late is 12 yuan. The comprehensive compensation cost assessment is 27 cost units.

[0093] For example, when the relaxation benefit value 42 is greater than the compensation cost value 27, a relaxation operation is performed on the quasi-rigid constraint. This means that when a candidate path is detected to violate a quasi-rigid or flexible constraint, the constraint is actively relaxed to maintain path feasibility. Specifically, the relaxation operation involves extending the arrival time of paths violating the customer time window constraint to a preset tolerance period; adjusting paths violating route preference constraints to alternative paths that are not preferred but are allowed; and rearranging the access order of stations that violate stop order constraints. Furthermore, the relaxation operation does not change the rigid constraint, and it must be verified before execution that the compensation cost corresponding to this operation is less than the loss incurred by not performing the operation.

[0094] Furthermore, when the relaxation benefit is less than or equal to the compensation cost, the relaxation operation on that constraint is abandoned, and the original candidate path remains unchanged, avoiding the introduction of unreasonable cost burdens or risk exposures in pursuit of local optimization. For example, if relaxing a certain elastic constraint can only bring a cost saving of 12 units of benefit, but requires bearing a customer complaint risk of 18 units of cost, then the relaxation is determined to be economically unreasonable, and the original path scheme is maintained.

[0095] Furthermore, when the conflict detection result contains at least one conflict, including quasi-rigid constraints or elastic constraints, and the relaxed path has been generated, the method further includes:

[0096] Perform a feasibility verification on the relaxed path to check whether the relaxed path violates any rigid constraint in the set of rigid constraints.

[0097] When the relaxed path violates any rigid constraint, the relaxed path is marked as an infeasible path and discarded, and the candidate path set is returned for reselection.

[0098] In this embodiment of the application, firstly, the feasibility of the relaxed path is verified to ensure that the constraint relaxation operation will not accidentally break through the rigid constraint bottom line. Since the relaxation operation of quasi-rigid constraints and elastic constraints involves changes such as path node adjustment, time window widening or road segment replacement, it may trigger a chain reaction, causing the rigid constraints such as vehicle load, driving range or legal driving time that were originally satisfied to be indirectly violated.

[0099] The specific verification process includes: recalculating the total mileage and estimated time of the relaxed route, and verifying whether the updated vehicle's remaining range covers the entire route; verifying the arrival times of each node after adjustment to ensure that the driver's continuous driving time does not exceed the legal limit of 4 hours and that rest intervals meet regulatory requirements; and checking whether the newly included road sections have traffic restrictions such as height restrictions, weight restrictions, or traffic prohibitions. For example, if a route changes from detouring through a national highway to crossing an urban expressway due to a relaxation of the customer's time window, although shortening the distance, the new road section prohibits trucks from passing during specific times. In this case, the relaxed route is deemed infeasible and discarded because it violates the traffic prohibition regulations in the rigid constraints.

[0100] Once the relaxed path passes the feasibility verification, it is included in the updated candidate path set, and the corresponding compensation cost value is associated as an additional attribute of the path scheme. If the relaxed path fails the verification, it returns to the original candidate path set, and other alternative paths that do not have quasi-rigid or elastic constraint conflicts are selected first, or the constraint relaxation evaluation is re-executed for candidate paths with a less severe conflict, until a feasible scheme that satisfies all rigid constraints is obtained or the candidate path set is exhausted.

[0101] S70: Based on the dynamic weight coefficient, the conflict detection result, and the compensation cost, select the optimal path scheme from the candidate path set or the relaxed path and output it.

[0102] In this embodiment, the optimal output path solution is obtained by adopting a comprehensive evaluation mechanism that integrates multi-attribute decision-making and dynamic preference, which transforms the path selection problem into a weighted multi-objective ranking problem.

[0103] Specifically, step S70 in the method includes:

[0104] When there are no constraint conflicts in the conflict detection results, each candidate path in the candidate path set is weighted and scored according to the dynamic weight coefficient, and the candidate path with the highest weighted score is selected as the optimal path solution output.

[0105] When the conflict detection results contain at least one conflict, including quasi-rigid constraints or elastic constraints, and the relaxed path has been generated, the comprehensive score of the relaxed path is calculated based on the dynamic weight coefficient and the compensation cost corresponding to the relaxed path, and the path with the highest comprehensive score is selected as the optimal path solution output.

[0106] The optimal route plan is sent to the vehicle terminal or navigation device in real time.

[0107] In this embodiment, firstly, when all paths in the candidate path set pass the rigid constraint test and do not trigger conflicts of quasi-rigid or elastic constraints, a weighted scoring mechanism based on dynamic weight coefficients is directly activated. The dynamic weight coefficients are generated driven by real-time business scenarios. For example, when a cold chain drug delivery task enters its final 2-hour countdown, the timeliness weight coefficient is automatically increased to 0.6, while the cost and carbon emission weights are correspondingly decreased to 0.25 and 0.15, respectively. The weighted scoring calculation formula at this time is: The one with the highest score is determined as the optimal path.

[0108] Secondly, when constraint conflicts exist and a relaxed path has been generated after relaxation operations, the comprehensive score calculation must incorporate the impact of compensation cost adjustments. Specifically, The compensation cost conversion factor is dynamically determined based on the current weight configuration. If cost weight dominates, the conversion factor is set to 0.8 to enhance cost sensitivity; if timeliness weight takes precedence, the conversion factor is reduced to 0.4 to tolerate moderate costs in exchange for time gains. For example, if a relaxed path has a basic weighted score of 87 points and an associated compensation cost of 23 units, the comprehensive score would be adjusted to [value missing] in a cost-sensitive scenario. In time-sensitive scenarios, the score is... The sorting results may be reversed in different scenarios.

[0109] Furthermore, once the optimal route is generated, it is sent in real time to the vehicle's onboard terminal or the driver's mobile terminal navigation device to guide actual driving operations.

[0110] Further, based on the dynamic weight coefficients, the conflict detection results, and the compensation cost, an optimal path scheme is selected and output from the candidate path set or the relaxed path, followed by:

[0111] The optimal path scheme is sent to the vehicle terminal for execution, and real-time multi-source perception data is continuously collected during the execution process;

[0112] When the deviation between the real-time multi-source sensing data during the execution process and the real-time multi-source sensing data on which the optimal path scheme is based exceeds a preset threshold, the step of loading the real-time multi-source sensing data of the transportation environment is triggered to the step of selecting the optimal path scheme, so as to obtain the updated optimal path scheme.

[0113] In this embodiment, after the optimal route plan is sent to the vehicle terminal, it enters the closed-loop monitoring and dynamic re-optimization stage. The vehicle IoT device continuously collects real-time multi-source perception data during the execution process, including GPS positioning coordinates, engine operating parameters, cabin temperature and humidity readings, driver physiological state monitoring data, and radar perception information of the surrounding traffic environment, forming an execution data stream that is aligned with the timing of the planning scheme.

[0114] Specifically, the deviation detection of real-time multi-source sensing data adopts a multi-dimensional threshold judgment mechanism. In the spatial dimension, a position deviation alarm is triggered when the lateral deviation of the actual vehicle trajectory from the planned path exceeds 200 meters or when it enters an unplanned road section for more than 500 meters. In the temporal dimension, a timeliness deviation alarm is triggered when the actual arrival time at each node deviates from the planned time by more than 15 minutes. In the environmental dimension, an environmental change alarm is triggered when the temperature control range of the vehicle compartment exceeds the ±2℃ range required for drug storage and transportation or when a traffic accident on the road ahead causes a traffic interruption. In the status dimension, a status abnormality alarm is triggered when the driver fatigue monitoring index exceeds the critical value or when an abnormal vehicle fault code is reported. Any deviation exceeding the preset threshold in any dimension is judged as a significant deviation, and a re-optimization process is initiated.

[0115] Furthermore, the triggering and execution of the re-optimization process follows the principle of combining rapid response with gradual correction. First, the subsequent node instructions of the current route plan are frozen, and a "route recalculation in progress" prompt message is pushed to the driver. At the same time, the complete optimization loop starting from step S10 is started in parallel: reloading the real-time multi-source perception data of the transportation environment, updating the status of the vehicle's remaining available resources, re-executing candidate route generation and constraint verification with the current vehicle position as the new starting point, generating the updated optimal route plan within 3 to 5 seconds and issuing instructions to cover the original instructions.

[0116] For example, a cold chain transport vehicle encounters a sudden closure of a bridge at the third node of its original planned route. After detecting that the route is impassable, it completes the recalculation of the detour plan within 2.8 seconds. The new plan increases the driving distance by 12 kilometers but ensures that the timeliness target deviation is controlled within 8 minutes. Simultaneously, the pre-cooling scheduling instructions of subsequent nodes are adjusted to maintain temperature control compliance.

[0117] In summary, compared with existing technologies, this application achieves layered control over the absolute safety baseline and flexible business needs in transportation route optimization by constructing a three-level constraint system of rigid constraints, quasi-rigid constraints, and flexible constraints. By introducing a quantitative evaluation mechanism for constraint relaxation and compensation costs, it provides an economically rational decision-making basis for reasonable adaptation of quasi-rigid constraints and flexible constraints while ensuring that rigid constraints cannot be infringed. Based on a multi-attribute decision-making framework with dynamic weight coefficients, route selection can respond in real time to the shift in target preferences under different transportation scenarios such as cold chain pharmaceuticals and hazardous chemicals, solving the technical defect that static weight configuration is difficult to adapt to dynamic business changes. Relying on the real-time perception and closed-loop re-optimization mechanism of the Internet of Things, route optimization is upgraded from "one-time planning" to "continuous iterative optimization", effectively addressing the problems of sudden changes in traffic environment and abnormal vehicle status during transportation execution.

[0118] In summary, the embodiments of this application have at least the following technical effects:

[0119] This application provides an IoT-based transportation route optimization method. First, by dynamically loading and processing real-time multi-source sensing data, a comprehensive digital mapping of the transportation environment is achieved, enabling route planning to make decisions based on the latest road conditions, vehicle status, and task requirements. Second, an environmental feature-driven multi-objective weight adaptive adjustment mechanism is adopted, overcoming the limitations of traditional fixed weights or manually set weights. This allows the priority of objectives such as cost, timeliness, and carbon emissions to be intelligently adjusted according to environmental changes, improving the environmental adaptability of multi-objective optimization. Third, through a hierarchical processing strategy of rigid constraints, quasi-rigid constraints, and flexible constraints, the relaxation benefits and compensation costs of adjustable constraints are quantitatively evaluated under the premise of ensuring traffic safety and regulatory compliance, expanding the feasible solution space and avoiding the loss of optimization opportunities due to simply abandoning conflicting paths. Finally, by combining Pareto optimality solutions and constraint relaxation mechanisms, a balance between global optimality and local flexibility is achieved. Through a comprehensive evaluation of dynamic weight coefficients, conflict detection results, and compensation costs, the output optimal route solution is ensured to be both economical and feasible in complex and ever-changing real-world transportation environments.

[0120] Through the above technical solutions, this application achieves intelligent, dynamic and refined transportation route optimization based on the fusion processing of real-time multi-source sensing data, adaptive dynamic adjustment of multi-objective weights, refined management of hierarchical constraints, and quantitative evaluation of constraint relaxation and compensation costs, effectively improving the decision-making quality and execution effect in complex transportation environments.

[0121] Example 2, as Figure 2 As shown, based on the same inventive concept as the IoT-based transportation route optimization method provided in Embodiment 1, this application also provides an IoT-based transportation route optimization system, including:

[0122] The data acquisition module 11 is used to load real-time multi-source sensing data of the transportation environment and obtain real-time multi-source sensing data, wherein the real-time multi-source sensing data includes road condition dynamic data, vehicle status data and task constraint data.

[0123] Function construction module 12 is used to construct a multi-objective dynamic optimization function based on the real-time multi-source sensing data, and obtain the multi-objective dynamic optimization function, wherein the multi-objective dynamic optimization function includes cost target, timeliness target and carbon emission target;

[0124] Feature extraction module 13 is used to extract environmental features from the real-time multi-source sensing data, perform adaptive adjustment of multi-objective weights based on the environmental features, and obtain the dynamic weight coefficients at the current moment.

[0125] Path optimization module 14 is used to perform Pareto optimal path solving based on the dynamic weight coefficients and the multi-objective dynamic optimization function to obtain a set of candidate paths;

[0126] The constraint condition acquisition module 15 is used to load the constraint condition set of the transportation environment and obtain the constraint condition set, wherein the constraint condition set includes rigid constraints, quasi-rigid constraints and elastic constraints.

[0127] The path detection module 16 is used to detect the conflict relationship between each candidate path in the candidate path set and the constraint condition set, and obtain the conflict detection result; when there is at least one conflict between quasi-rigid constraints and elastic constraints in the conflict detection result, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost.

[0128] The scheme acquisition module 17 is used to select the optimal path scheme from the candidate path set or the relaxed path based on the dynamic weight coefficient, the conflict detection result and the compensation cost.

[0129] In one embodiment, the data acquisition module 11 is specifically used for:

[0130] Obtain the dynamic traffic data, which includes real-time congestion index, average traffic speed, traffic accident information, and weather conditions;

[0131] Obtain the vehicle status data, wherein the vehicle status data includes vehicle location, remaining driving range, load status and driving time;

[0132] Obtain the task constraint data, which includes customer time window, order urgency, and regional environmental protection requirement level.

[0133] Furthermore, in one embodiment, environmental features are extracted from the real-time multi-source sensing data, and adaptive adjustment of multi-objective weights is performed based on the environmental features to obtain the dynamic weight coefficients at the current moment, which includes the following steps:

[0134] Collect historical environmental feature sequences and historical optimal weight sequences under preset transportation scenarios, wherein the historical environmental feature sequences include historical congestion index sequences, historical order urgency sequences, and historical weather level sequences;

[0135] Using the historical environmental feature sequence as input and the historical optimal weight sequence as supervision, a weight adaptive adjustment model is trained to obtain the weight adaptive adjustment model, wherein the weight adaptive adjustment model adopts a fuzzy neural network or a deep reinforcement learning network.

[0136] The weight adaptive adjustment model is associated with and stored with the preset transportation scenario, and added to the weight adjustment model library;

[0137] Based on the current transportation scenario, the adaptive weight adjustment model in the weight adjustment model library is invoked to process the environmental features in the real-time multi-source perception data and output the dynamic weight coefficient at the current moment.

[0138] Further, in one embodiment, detecting the conflict relationship between each candidate path in the candidate path set and the constraint set to obtain conflict detection results includes:

[0139] Load the rigid constraints to obtain a set of rigid constraints, wherein the rigid constraints include traffic rule speed limits, restricted areas, and bridge load-bearing limits;

[0140] When any candidate path violates any rigid constraint in the set of rigid constraints, the candidate path is marked as an infeasible path and removed from the set of candidate paths to obtain an updated set of candidate paths.

[0141] Load the quasi-rigid constraints to obtain a set of quasi-rigid constraints, wherein the quasi-rigid constraints include customer time windows and driving duration regulations;

[0142] Load the elastic constraints to obtain a set of elastic constraints, wherein the elastic constraints include route preferences and stop order;

[0143] The degree of deviation between each candidate path in the updated candidate path set and the quasi-rigid constraint set and the elastic constraint set is recorded as the conflict detection result.

[0144] Furthermore, when at least one of the quasi-rigid constraints and elastic constraints exists in the conflict detection results, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost, including:

[0145] Extract the quasi-rigid or elastic constraints that have conflicted from the conflict detection results and set them as constraints to be relaxed.

[0146] Calculate the relaxation benefit of the constraint to be relaxed and obtain the relaxation benefit value. The calculation basis of the relaxation benefit includes time saving, energy consumption reduction and task completion rate improvement.

[0147] Calculate the compensation cost of the constraint to be relaxed and obtain the compensation cost value. The calculation basis of the compensation cost includes time window delay penalty, detour additional cost, and additional cost of change of transportation mode.

[0148] When the relaxation benefit value is greater than the compensation cost value, a relaxation operation is performed on the constraint to be relaxed, a relaxed path is generated, and the compensation cost value is associated with the relaxed path.

[0149] When the relaxation benefit value is less than or equal to the compensation cost value, the relaxation operation is abandoned and the original candidate path is maintained.

[0150] In one embodiment, the solution acquisition module 17 is specifically used for:

[0151] When there are no constraint conflicts in the conflict detection results, each candidate path in the candidate path set is weighted and scored according to the dynamic weight coefficient, and the candidate path with the highest weighted score is selected as the optimal path solution output.

[0152] When the conflict detection results contain at least one conflict, including quasi-rigid constraints or elastic constraints, and the relaxed path has been generated, the comprehensive score of the relaxed path is calculated based on the dynamic weight coefficient and the compensation cost corresponding to the relaxed path, and the path with the highest comprehensive score is selected as the optimal path solution output.

[0153] The optimal route plan is sent to the vehicle terminal or navigation device in real time.

[0154] Further, based on the dynamic weight coefficients, the conflict detection results, and the compensation cost, an optimal path scheme is selected and output from the candidate path set or the relaxed path, followed by:

[0155] The optimal path scheme is sent to the vehicle terminal for execution, and real-time multi-source perception data is continuously collected during the execution process;

[0156] When the deviation between the real-time multi-source sensing data during the execution process and the real-time multi-source sensing data on which the optimal path scheme is based exceeds a preset threshold, the step of loading the real-time multi-source sensing data of the transportation environment is triggered to the step of selecting the optimal path scheme, so as to obtain the updated optimal path scheme.

[0157] Furthermore, based on the environmental features in the real-time multi-source sensing data, adaptive adjustment of multi-objective weights is performed to obtain the dynamic weight coefficients at the current moment, followed by:

[0158] The dynamic weight coefficients at the current time and the previous time are smoothed by exponential moving average to obtain smoothed dynamic weight coefficients, and the smoothed dynamic weight coefficients are used as the updated dynamic weight coefficients at the current time.

[0159] Furthermore, when the conflict detection result contains at least one conflict, including quasi-rigid constraints or elastic constraints, and the relaxed path has been generated, the method further includes:

[0160] Perform a feasibility verification on the relaxed path to check whether the relaxed path violates any rigid constraint in the set of rigid constraints.

[0161] When the relaxed path violates any rigid constraint, the relaxed path is marked as an infeasible path and discarded, and the candidate path set is returned for reselection.

Claims

1. A method for optimizing a transportation route based on Internet of Things, characterized in that, include: Load real-time multi-source perception data of the transportation environment to obtain real-time multi-source perception data, wherein the real-time multi-source perception data includes road condition dynamic data, vehicle status data and task constraint data; Based on the real-time multi-source sensing data, a multi-objective dynamic optimization function is constructed to obtain the multi-objective dynamic optimization function, wherein the multi-objective dynamic optimization function includes cost target, timeliness target and carbon emission target; Environmental features are extracted from the real-time multi-source sensing data, and adaptive adjustment of multi-objective weights is performed based on the environmental features to obtain the dynamic weight coefficients at the current moment. Based on the dynamic weighting coefficients and the multi-objective dynamic optimization function, Pareto optimal path solving is performed to obtain a candidate path set; Load the constraint set of the transportation environment to obtain the constraint set, wherein the constraint set includes rigid constraints, quasi-rigid constraints and elastic constraints; The conflict relationship between each candidate path in the candidate path set and the constraint condition set is detected to obtain the conflict detection result; when there is at least one conflict between quasi-rigid constraints and elastic constraints in the conflict detection result, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost. Based on the dynamic weight coefficients, the conflict detection results, and the compensation cost, the optimal path scheme is selected and output from the candidate path set or the relaxed path. 2.The Internet of Things based transportation route optimization method of claim 1, wherein, Load real-time multi-source sensing data of the transportation environment to obtain real-time multi-source sensing data, including: Obtain the dynamic traffic data, which includes real-time congestion index, average traffic speed, traffic accident information, and weather conditions; Obtain the vehicle status data, wherein the vehicle status data includes vehicle location, remaining driving range, load status and driving time; Obtain the task constraint data, which includes customer time window, order urgency, and regional environmental protection requirement level. 3.The Internet of Things based transportation route optimization method of claim 1, wherein, Environmental features are extracted from the real-time multi-source sensing data, and adaptive adjustment of multi-objective weights is performed based on the environmental features to obtain the dynamic weight coefficients at the current moment. This process includes: Collect historical environmental feature sequences and historical optimal weight sequences under preset transportation scenarios, wherein the historical environmental feature sequences include historical congestion index sequences, historical order urgency sequences, and historical weather level sequences; Using the historical environmental feature sequence as input and the historical optimal weight sequence as supervision, a weight adaptive adjustment model is trained to obtain the weight adaptive adjustment model, wherein the weight adaptive adjustment model adopts a fuzzy neural network or a deep reinforcement learning network. The weight adaptive adjustment model is associated with and stored with the preset transportation scenario, and added to the weight adjustment model library; Based on the current transportation scenario, the adaptive weight adjustment model in the weight adjustment model library is invoked to process the environmental features in the real-time multi-source perception data and output the dynamic weight coefficient at the current moment.

4. The transportation route optimization method based on the Internet of Things as described in claim 1, characterized in that, Detecting the conflict relationship between each candidate path in the candidate path set and the constraint set, and obtaining the conflict detection result, including: Load the rigid constraints to obtain a set of rigid constraints, wherein the rigid constraints include traffic rule speed limits, restricted areas, and bridge load-bearing limits; When any candidate path violates any rigid constraint in the set of rigid constraints, the candidate path is marked as an infeasible path and removed from the set of candidate paths to obtain an updated set of candidate paths. Load the quasi-rigid constraints to obtain a set of quasi-rigid constraints, wherein the quasi-rigid constraints include customer time windows and driving duration regulations; Load the elastic constraints to obtain a set of elastic constraints, wherein the elastic constraints include route preferences and stop order; The degree of deviation between each candidate path in the updated candidate path set and the quasi-rigid constraint set and the elastic constraint set is recorded as the conflict detection result.

5. The transportation route optimization method based on the Internet of Things as described in claim 1, characterized in that, When the conflict detection results contain at least one conflict between quasi-rigid constraints and elastic constraints, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost, including: Extract the quasi-rigid or elastic constraints that have conflicted from the conflict detection results and set them as constraints to be relaxed. Calculate the relaxation benefit of the constraint to be relaxed and obtain the relaxation benefit value. The calculation basis of the relaxation benefit includes time saving, energy consumption reduction and task completion rate improvement. Calculate the compensation cost of the constraint to be relaxed and obtain the compensation cost value. The calculation basis of the compensation cost includes time window delay penalty, detour additional cost, and additional cost of change of transportation mode. When the relaxation benefit value is greater than the compensation cost value, a relaxation operation is performed on the constraint to be relaxed, a relaxed path is generated, and the compensation cost value is associated with the relaxed path. When the relaxation benefit value is less than or equal to the compensation cost value, the relaxation operation is abandoned and the original candidate path is maintained.

6. The transportation route optimization method based on the Internet of Things as described in claim 1, characterized in that, Based on the dynamic weight coefficients, the conflict detection results, and the compensation cost, the optimal path scheme is selected and output from the candidate path set or the relaxed path, including: When there are no constraint conflicts in the conflict detection results, each candidate path in the candidate path set is weighted and scored according to the dynamic weight coefficient, and the candidate path with the highest weighted score is selected as the optimal path solution output. When the conflict detection results contain at least one conflict, including quasi-rigid constraints or elastic constraints, and the relaxed path has been generated, the comprehensive score of the relaxed path is calculated based on the dynamic weight coefficient and the compensation cost corresponding to the relaxed path, and the path with the highest comprehensive score is selected as the optimal path solution output. The optimal route plan is sent to the vehicle terminal or navigation device in real time.

7. The transportation route optimization method based on the Internet of Things as described in claim 1, characterized in that, Based on the dynamic weight coefficients, the conflict detection results, and the compensation cost, the optimal path scheme is selected from the candidate path set or the relaxed path and output, followed by: The optimal path scheme is sent to the vehicle terminal for execution, and real-time multi-source perception data is continuously collected during the execution process; When the deviation between the real-time multi-source sensing data during the execution process and the real-time multi-source sensing data on which the optimal path scheme is based exceeds a preset threshold, the step of loading the real-time multi-source sensing data of the transportation environment is triggered to the step of selecting the optimal path scheme, so as to obtain the updated optimal path scheme.

8. The transportation route optimization method based on the Internet of Things as described in claim 1, characterized in that, Based on the environmental features in the real-time multi-source sensing data, adaptive adjustment of multi-objective weights is performed to obtain the dynamic weight coefficients at the current moment, followed by: The dynamic weight coefficients at the current time and the previous time are smoothed by exponential moving average to obtain smoothed dynamic weight coefficients, and the smoothed dynamic weight coefficients are used as the updated dynamic weight coefficients at the current time.

9. The transportation route optimization method based on the Internet of Things as described in claim 1, characterized in that, When the conflict detection result contains at least one conflict, including quasi-rigid constraints or elastic constraints, and the relaxed path has been generated, the method further includes: Perform a feasibility verification on the relaxed path to check whether the relaxed path violates any rigid constraint in the set of rigid constraints. When the relaxed path violates any rigid constraint, the relaxed path is marked as an infeasible path and discarded, and the candidate path set is returned for reselection.

10. A transportation route optimization system based on the Internet of Things, characterized in that, The method for executing the Internet of Things-based transportation route optimization method according to any one of claims 1-9 includes: The data acquisition module is used to load real-time multi-source sensing data of the transportation environment and obtain real-time multi-source sensing data, wherein the real-time multi-source sensing data includes road condition dynamic data, vehicle status data and task constraint data; The function construction module is used to construct a multi-objective dynamic optimization function based on the real-time multi-source sensing data, and obtain the multi-objective dynamic optimization function, wherein the multi-objective dynamic optimization function includes cost target, timeliness target and carbon emission target; The feature extraction module is used to extract environmental features from the real-time multi-source sensing data, perform adaptive adjustment of multi-objective weights based on the environmental features, and obtain the dynamic weight coefficients at the current moment. The path optimization module is used to perform Pareto optimal path solving based on the dynamic weight coefficients and the multi-objective dynamic optimization function to obtain a set of candidate paths. The constraint condition acquisition module is used to load the constraint condition set of the transportation environment and obtain the constraint condition set, wherein the constraint condition set includes rigid constraints, quasi-rigid constraints and elastic constraints; The path detection module is used to detect the conflict relationship between each candidate path in the candidate path set and the constraint condition set, and obtain the conflict detection result; when there is at least one conflict between quasi-rigid constraints and elastic constraints in the conflict detection result, constraint relaxation and compensation cost calculation are performed to obtain the relaxed path and the corresponding compensation cost. The scheme acquisition module is used to select the optimal path scheme from the candidate path set or the relaxed path based on the dynamic weight coefficient, the conflict detection result and the compensation cost.

Images