A big data-based renewable resource recycling system and method

Abstract

The application relates to the field of resource recycling treatment, and discloses a renewable resource recycling system and method based on big data, which comprises a data acquisition module, a resource recycling analysis module and a planning and scheduling module. Through multi-source data acquisition, structured data sets are formed after preprocessing and fusion. The energy cost is accurately calculated based on a vehicle dynamics model, the service urgency cost is quantified through a regional gridding model, and the real-time traffic estimation time cost is combined. The global state of the system is predicted by analyzing the time sequence slope sequence characteristics. According to the system state, the adaptive weight model is used to dynamically generate the weight of each cost, a weighted comprehensive cost function is constructed, and a path optimization algorithm is used to solve the optimal collection and transportation path sequence under the conditions of meeting the load and task constraints. Finally, the path sequence is decomposed into operation instructions and issued to the vehicle terminal. The application realizes the leap from static planning to dynamic adaptive scheduling, and improves the economy, efficiency and service quality of resource recycling operations.

Description

Technical Field

[0001] This invention relates to the field of resource recycling and processing, and in particular to a big data-based recycling system and method for renewable resources. Background Technology

[0002] The efficient recycling and utilization of renewable resources is a key link in promoting the development of a circular economy and achieving sustainable urban operation. It is not only related to resource conservation and environmental protection, but also has crucial strategic significance for the reduction and harmless treatment of urban waste. With the improvement of social consumption levels, the amount of renewable resources generated is increasing daily, placing higher demands on the efficiency and intelligence level of its recycling system.

[0003] Currently, the recycling and operation of renewable resources mainly relies on two models: one is a planned collection model with fixed routes and fixed cycles, in which recycling vehicles collect resources along preset routes; the other is to receive orders via telephone or a simple information platform, with dispatchers assigning orders and arranging routes based on experience. These methods have achieved initial resource collection to some extent, but they are all passive and extensive management methods.

[0004] However, the existing methods mentioned above have obvious shortcomings and defects. The fixed route model is rigid and cannot respond to dynamic changes in the inventory of recycling points and sudden task demands, which can easily lead to problems such as high vehicle empty load rate and failure to clear fully loaded recycling points in a timely manner. On the other hand, the scheduling method that relies on human experience is slow and subjective in the decision-making process, and it is difficult to comprehensively optimize multiple objectives such as vehicle energy consumption, time cost and service response speed. Especially in large-scale recycling networks, the existing technology lacks the ability to dynamically perceive and quantify key factors such as real-time traffic, vehicle load and regional order backlog risk, resulting in unscientific route planning, high operating costs, and bottlenecks in overall recycling efficiency and service quality. Summary of the Invention

[0005] The objective of this invention can be achieved through the following technical solution: a big data-based renewable resource recycling system, comprising:

[0006] The data acquisition module obtains static basic information about the recycling network and vehicle operating parameters through the system configuration database, IoT sensors deployed at recycling nodes and vehicles, user terminal interfaces, and third-party map and energy service APIs. It also performs standardized preprocessing on the collected raw data and finally aggregates the multi-source data into a structured fusion dataset according to node number and time index.

[0007] The resource recycling analysis module calculates the estimated total energy consumption based on vehicle dynamics models and real-time parameters, and calculates energy costs by combining energy unit prices. Simultaneously, it calculates node urgency based on gridded areas, historical data, and order waiting times, and defines regional urgency accordingly. It also estimates the total task time based on real-time traffic and node workload, and calculates the final time cost by combining a unit time cost coefficient. Finally, it extracts feature vectors by analyzing the time-series slope sequence fitted by regional urgency and order waiting times, and outputs system state values ​​and specific states using a pre-built classifier.

[0008] The planning and scheduling module quantifies the received system state values ​​into specific numerical values ​​and inputs them, along with preset strategy bias coefficients, into an adaptive weight calculation model based on exponential functions and normalization processing to dynamically generate the weights of various costs. Subsequently, an optimization function is constructed with the goal of minimizing the weighted comprehensive cost, and the solution is obtained based on a path optimization algorithm under the conditions of satisfying vehicle load, task uniqueness, and capacity constraints, outputting the optimal collection route sequence for the fleet. Finally, the route sequence is decomposed into navigation and operation instructions and sent to the vehicle terminals through a wireless communication network.

[0009] Preferably, the static basic information and vehicle operating parameters specifically include:

[0010] Static basic information includes unique identifiers for all resource recycling nodes, GPS latitude and longitude coordinates and the theoretical maximum capacity of the recycling nodes, unique identifiers for all resource recycling vehicles, the energy type of the vehicles, and the reference rolling resistance coefficient and air resistance coefficient of the vehicles;

[0011] Vehicle operating parameters include real-time vehicle load capacity, rated load capacity, empty vehicle weight, equipment full load status indicator, real-time location, driving speed, remaining load capacity, vehicle battery power or fuel balance uploaded by the resource recycling node.

[0012] Preferably, the estimated total energy consumption specifically refers to:

[0013] Based on dynamic parameters consisting of real-time vehicle load, road gradient, and driving speed, an energy consumption calculation model is used to estimate the energy consumption of the vehicle on the road segment, rather than simply multiplying the driving distance by a fixed energy consumption coefficient. The energy consumption of all road segments is then summed to obtain the estimated total energy consumption to complete the current task.

[0014] Preferably, the urgency of the region is specifically as follows:

[0015] The recycling service area is divided into multiple regular grids on the electronic map. For each grid with unprocessed orders, the node urgency of the grid is calculated. The node urgency of all grids with unprocessed orders that are not covered by the candidate path is obtained. The sum of the node urgency of all such grids in the area is the area urgency.

[0016] Preferably, the final time cost is specifically:

[0017] The candidate path is divided into several segments. For each segment, the real-time average driving speed and the length of the segment are obtained, and the estimated travel time on the segment is calculated.

[0018] For each recycling node that needs to stop along the route, the operation time at that node is estimated based on the real-time inventory weight collected at that node and a preset unit weight operation efficiency; the total time is then the sum of the road segment travel time and the node operation time.

[0019] Based on the set unit time cost coefficient, the total time is multiplied by the unit time cost coefficient to obtain the final time cost of executing the candidate path.

[0020] Preferably, the system state value is specifically:

[0021] A linear regression was performed with the regional urgency value of each network as the X-axis and the average waiting time of all unprocessed orders within that network as the Y-axis to obtain the slope of the trend line;

[0022] The system iteratively performs the above fitting process within a fixed time window to obtain a time-series slope sequence;

[0023] Feature extraction is performed on the time-series slope sequence to obtain a feature vector containing a positive intensity index, a negative recovery index, a stability index, and a trend persistence index;

[0024] The feature vector is input into a preset classifier, which outputs the system state value.

[0025] Preferably, the weights of the various costs are as follows:

[0026] The system state identifier is quantized into a system state quantization value;

[0027] The quantified values ​​of the system state and the preset strategy bias coefficients corresponding to each cost are input into an adaptive weight calculation model;

[0028] The adaptive weight calculation model is configured to: use the product of the system state quantization value and the bias coefficients of each strategy as the input of the natural exponential function, and normalize the output values ​​of each exponential function to obtain the dynamic weights of each cost.

[0029] Preferably, the optimal collection and transportation route sequence is as follows:

[0030] The comprehensive cost function is calculated by weighting the regional urgency cost, energy cost, and time cost. The metaheuristic algorithm is used to solve the problem under the conditions of satisfying vehicle load constraints and task uniqueness constraints, and the optimal collection and transportation route sequence is output.

[0031] Preferably, the step of decomposing the optimal collection and transportation route sequence into navigation and operation instructions specifically includes:

[0032] The optimal collection route sequence is parsed into a series of continuous navigation segments and docking point operation instructions based on the order of nodes along the route and the corresponding operation tasks of each node;

[0033] The navigation segment and docking point operation instructions are packaged and distributed to the corresponding vehicle terminal or the driver's mobile application via wireless communication network.

[0034] A preferred method for recycling renewable resources based on big data includes the following steps:

[0035] S1. Through various sensing devices and data interfaces deployed in the recycling network, real-time static basic information and vehicle operating parameters related to the recycling of renewable resources are acquired; and standardized preprocessing is performed on the collected raw data, and finally, the multi-source data is aggregated into a structured fusion dataset according to node number and time index;

[0036] S2. Based on the vehicle dynamics model and real-time parameters, the total energy consumption is estimated, and the energy cost is calculated in conjunction with the energy unit price. Simultaneously, based on the gridded area, historical data, and order waiting time, the urgency of nodes is calculated, and the area urgency cost is defined accordingly. Furthermore, the total task time is estimated based on real-time traffic and node workload, and the time cost is calculated in conjunction with the unit time cost coefficient. Finally, by analyzing the time-series slope sequence fitted by area urgency and order waiting time, feature vectors are extracted, and a pre-set classifier is used to output the system state value and specific state.

[0037] S3 is used to receive system status identifiers, quantify them into specific values, and input them together with preset strategy bias coefficients into an adaptive weight calculation model based on exponential functions and normalization processing to dynamically generate the weights of various costs. Subsequently, an optimization function with the goal of minimizing the weighted comprehensive cost is constructed, and the solution is performed based on the path optimization algorithm under the conditions of satisfying vehicle load, task uniqueness, and capacity constraints, outputting the optimal collection route sequence of the fleet. Finally, the optimal collection route sequence is decomposed into navigation and operation instructions, which are sent to the vehicle terminals through the wireless communication network.

[0038] The beneficial effects of this invention:

[0039] (1) By introducing a vehicle dynamics model for energy consumption calculation, this invention overcomes the huge error of traditional simple mileage estimation and achieves accurate prediction of fuel / electricity consumption, providing a reliable basis for energy saving and cost reduction. At the same time, the accurate prediction of driving and operating time makes time cost accounting more realistic, effectively reducing vehicle idleness and ineffective driving, and significantly improving overall operating efficiency.

[0040] (2) By dynamically modeling multiple cost factors, including energy costs, task urgency, time costs, and path load distribution, and combining normalization processing with a dynamic weight-driven strategy planning algorithm, this invention can generate optimal collection and transportation routes that better meet real-time operational constraints. This results in reduced vehicle mileage, optimized attendance frequency, and rationalized resource allocation, thereby effectively reducing overall operating costs and improving resource recovery efficiency.

[0041] (3) By transforming service delay risk into quantifiable regional urgency costs and incorporating them into path optimization objectives based on a dynamic weight model, the timeliness of service response and the accuracy of resource scheduling are effectively improved. Through grid-based monitoring and time-series trend prediction, the system can detect areas where service is about to become unbalanced in advance and prioritize their scheduling in path planning through cost weights. This fundamentally changes the situation of passively responding to complaints, achieves proactive and forward-looking service quality assurance, and improves customer satisfaction. Attached Figure Description

[0042] The invention will now be further described with reference to the accompanying drawings.

[0043] Figure 1 This is a schematic diagram of a big data-based recycling system for renewable resources according to the present invention.

[0044] Figure 2 This is a flowchart of a big data-based recycling method for renewable resources according to the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Taking a recycling company as an example, the company is responsible for the intelligent collection and transportation of recyclables covering multiple communities and supermarkets throughout the city. During operation, the system collects multi-dimensional data in real time, including vehicle GPS trajectory, user delivery frequency, recycling bin overflow rate, energy consumption, battery life, operation timeliness, and road congestion index, and processes the data collaboratively with the cloud big data platform through edge devices.

[0047] To resolve the above technical issues, please refer to [link / reference]. Figure 1 As shown, this invention is a big data-based recycling system for renewable resources, comprising a data acquisition module, a resource recycling analysis module, and a planning and scheduling module;

[0048] The data acquisition module obtains static basic information about the recycling network and vehicle operating parameters through the system configuration database, IoT sensors deployed at recycling nodes and vehicles, user terminal interfaces, and third-party map and energy service APIs. It also performs standardized preprocessing on the collected raw data and finally aggregates the multi-source data into a structured fusion dataset according to node number and time index.

[0049] The resource recycling analysis module calculates the estimated total energy consumption based on vehicle dynamics models and real-time parameters, and calculates energy costs by combining energy unit prices. Simultaneously, it calculates node urgency based on gridded areas, historical data, and order waiting times, and defines regional urgency accordingly. It also estimates the total task time based on real-time traffic and node workload, and calculates time costs by combining unit time cost coefficients. Finally, it extracts feature vectors by analyzing the time-series slope sequence fitted by regional urgency and order waiting times, and outputs system state values ​​and specific states using a pre-built classifier.

[0050] The planning and scheduling module quantifies the received system state values ​​into specific numerical values ​​and inputs them, along with preset strategy bias coefficients, into an adaptive weight calculation model based on exponential functions and normalization to dynamically generate weights for various costs. Subsequently, an optimization function is constructed with the goal of minimizing the weighted overall cost, and a path optimization algorithm is used to solve the problem under constraints of vehicle load, task uniqueness, and capacity, outputting the optimal collection route sequence for the fleet. Finally, the optimal collection route sequence is decomposed into navigation and operational instructions, which are then transmitted to the vehicle terminals via a wireless communication network.

[0051] Furthermore, to better illustrate the technical solutions of the embodiments of the present invention, a big data-based renewable resource recycling system is provided, such as... Figure 2 As shown, this embodiment of the invention provides a method for recycling renewable resources based on big data, specifically including the following steps:

[0052] S1. Through various sensing devices and data interfaces deployed in the recycling network, real-time static basic information and vehicle operating parameters related to the recycling of renewable resources are acquired; and standardized preprocessing is performed on the collected raw data, and finally, the multi-source data is aggregated into a structured fusion dataset according to node number and time index;

[0053] S1 includes the following steps:

[0054] S11. The system obtains static basic information of the recycling network from a pre-set database or through a management configuration interface. The static basic information includes:

[0055] The unique identifier, GPS latitude and longitude coordinates, and theoretical maximum capacity of each resource recycling node (such as smart recycling bins, recycling stations, resource processing centers, or transfer stations);

[0056] The unique identifier for each resource recycling vehicle, the vehicle's energy type, and the vehicle's baseline rolling resistance coefficient and air resistance coefficient;

[0057] S12. Vehicle operating parameters of the recycling network's operational status are acquired in real time via IoT sensing devices, user terminal interfaces, and external data interfaces deployed at each recycling node; the vehicle operating parameters include:

[0058] The resource recycling nodes upload real-time vehicle load capacity, rated load capacity, empty vehicle weight, and equipment full load status indicators.

[0059] Real-time location, speed, remaining load capacity, battery charge, or fuel level of the recycling vehicle;

[0060] The external public interface provides meteorological data, traffic congestion index, order information, and resource recycling area types; resource recycling area types include commercial areas, industrial areas, and residential areas.

[0061] S13. The acquired static basic information and vehicle operating parameters are aggregated and uniformly encoded to form the original multi-source dataset, and preprocessing operations are performed on the original multi-source dataset, including:

[0062] Abnormal data detection and removal: eliminating invalid, redundant, and misformatted records;

[0063] Missing value imputation: missing data items are recovered using methods such as temporal interpolation, spatial proximity estimation, or historical mean imputation.

[0064] Timestamp alignment and spatial coordinate mapping unify data from different sampling periods to a set time step and perform spatial registration based on geographic coordinates;

[0065] Data standardization is performed because the physical dimensions and numerical ranges of different features in the multi-source datasets vary greatly (e.g., inventory weight is on the order of hundreds of kilograms, while traffic index is on the order of single digits). To avoid the difference in dimensions causing some features to be masked by their numerical magnitude during subsequent data fusion and feature weighting, this invention employs min-max normalization or Z-score standardization methods to transform all features to a uniform dimensionless scale. It should be noted that this standardization process is intended to serve the fusion and weighted analysis of multi-source data. For subsequent specialized calculations based on physical laws (such as the vehicle dynamics energy consumption calculation described in S21), the system will directly call its original parameter values ​​with physical units.

[0066] S14. Perform feature fusion on the standardized multi-source data, specifically as follows:

[0067] Data from different sources are aggregated by node number and time index, and then weighted and integrated based on the historical recycling volume, equipment status, and environmental characteristics of the recycling nodes according to the time window to form a structured fusion dataset;

[0068] The fused dataset serves as input data for subsequent recycling demand prediction models;

[0069] In this embodiment, through the above steps, the system achieves efficient collection and fusion of static network information and dynamic operating parameters, constructing a data foundation with a unified format, continuous time, and spatial alignment, providing reliable data support for subsequent predictive analysis and dynamic scheduling;

[0070] S2. By constructing a comprehensive analytical and predictive model, the execution costs of recycling tasks in three dimensions—energy, time, and service urgency—are quantitatively assessed, and the overall operational status of the system is diagnosed, providing accurate and multi-objective quantitative inputs for subsequent path optimization decisions.

[0071] S2 includes the following steps:

[0072] S21. Based on the acquired vehicle operating parameters, including the vehicle's real-time load capacity, rated load capacity, and empty vehicle weight, calculate the vehicle's real-time total mass;

[0073] Based on the acquired real-time average driving speed and road grade, the vehicle's baseline rolling resistance coefficient and air resistance coefficient are determined and pre-stored in the vehicle information database.

[0074] By calling the elevation service of a third-party map API, the elevation of each node along the planned route is obtained, and the slope value of the road segment is calculated accordingly.

[0075] Input the above parameters into the vehicle dynamics energy consumption model to calculate the energy consumption of the vehicle on road segment i:

[0076] Where Ei is the energy consumption (kJ or kWh) of the vehicle in road segment I, Di is the total length of road segment i, i = 1, 2, 3...I, i is a positive integer, i represents the sequence number of the road segment, and Fi is the rolling resistance. μ is the rolling resistance coefficient, m is the real-time total mass, g is the gravitational acceleration, θ is the slope angle, and Zi is the slope resistance. Qi is air resistance. Where c is the drag coefficient, A is the vehicle's frontal area, ρ is the air density, and V is the speed (m / s); the above vehicle dynamics energy consumption model is based on dynamic parameters consisting of the vehicle's real-time load mass, road gradient, and driving speed, and is estimated through an energy consumption calculation model, rather than simply based on the product of driving mileage and a fixed energy consumption coefficient;

[0077] Sum the energy consumption i of all road segments to obtain the estimated total energy consumption E for completing the current task. 总 ;

[0078] S22. The system obtains the real-time energy unit price P (e.g., fuel price in yuan / liter, or charging price in yuan / kWh) for the current region from the accessed energy price service API; and performs cost calculations based on the vehicle's energy type:

[0079] If the vehicle being recycled is a gasoline-powered vehicle, the system obtains the real-time fuel price H1 (unit: yuan / liter) through the energy price interface, and then calculates the estimated total energy consumption E. 总 Convert using the following formula: ,where L 燃油 The fuel equivalent per unit energy (unit: kWh / L) for the vehicle engine can be obtained from the vehicle manufacturer's calibration value or by fitting historical vehicle operating data. The final fuel energy cost calculation method is as follows: Finally, the fuel energy cost C1 is obtained;

[0080] If the recycled vehicle is a pure electric vehicle, the system obtains the real-time electricity price H2 (unit: yuan / kWh) and calculates the electricity cost according to the following formula: Ultimately, the cost of electricity, C2, is obtained.

[0081] If it is a hybrid vehicle, based on its pre-set fuel-electric energy distribution model, the preset fuel and electric energy ratio parameters are β1 and β2, and β1 + β2 = 1; then the hybrid energy cost is calculated as follows:

[0082] The final cost of hybrid energy is C3;

[0083] The energy cost assessment results for the vehicle performing the current task can be obtained through the above calculation method, providing a basis for subsequent analysis;

[0084] S23. Dynamically assess the order backlog risk in different geographical regions, and predict the regions that need priority scheduling through data analysis of global trends, thereby achieving proactive and optimized resource allocation; specifically including:

[0085] S231. Divide the recycling service area into multiple regular grids (such as 1km x 1km square grids) on an electronic map, and assign a unique area ID to each grid;

[0086] A static base urgency coefficient Bi is calculated for each grid cell. This coefficient is based on:

[0087] Historical daily average order density: The average number of orders generated within this grid over the past year;

[0088] Area type coefficient: A preset weighting coefficient, for example, commercial areas and upscale residential areas are assigned higher coefficients, while industrial areas and suburbs are assigned lower coefficients;

[0089] Real-time monitoring of all generated but unprocessed recycling orders, obtaining the grid ID of the recycling order, and the order's waiting time (in minutes) is calculated. The average waiting time is denoted as Ti.

[0090] S232. For each grid i with unprocessed orders, calculate its node urgency Ui. This value combines static basic attributes and dynamic waiting conditions, and the calculation formula is as follows:

[0091] α1 and α2 are preset normalized weighting coefficients used to balance the impact of historical trends and real-time emergency situations;

[0092] The regional urgency is defined as the sum of the urgency of nodes in all grids with unprocessed orders that are not covered by a candidate path. The sum of the node urgency of all grid i in the region is the regional urgency.

[0093] S233. Based on the collected data, predict the total time required for the vehicle to travel along the candidate path, where the total time includes the road segment travel time and the node operation time; specifically as follows:

[0094] Segment travel time: Divide the candidate route into several segments. For each segment, obtain the real-time average travel speed and the length of the segment, and calculate the estimated travel time on that segment.

[0095] Node operation time: For each recycling node that needs to stop on the path, the operation time at that node is estimated based on the real-time inventory weight collected at that node and a preset unit weight operation efficiency (in kg / s);

[0096] The total time is the sum of the travel time on the road segment and the operation time at each node;

[0097] A unit time cost coefficient is set, which is calculated based on the company's historical operating data and pre-stored in the system database for periodic updates. This coefficient is then used to analyze the company's labor costs, vehicle fixed depreciation, and management and operating costs.

[0098] Multiply the total time obtained by the unit time cost coefficient to get the final time cost of executing the candidate path, and denote it as Gi;

[0099] S234. Generate a data point in a two-dimensional Cartesian coordinate system with the regional urgency value of each network as the X-axis and the average waiting time of all unprocessed orders within that network as the Y-axis;

[0100] The least squares method is used to perform linear regression fitting on all grid data points to obtain a straight line that best represents the overall trend. Its equation is Y=kX+b, where k is the slope of the trend line.

[0101] The system executes steps S231 to S232 cyclically with a configurable fixed time window as the period, thereby obtaining a slope sequence K={k1,k2,k3,...,kn} arranged in chronological order, where n is the number of historical cycles counted. This sequence constitutes the time-series slope sequence of the system state.

[0102] The sampling frequency is set as follows: the system automatically adjusts the sampling frequency based on the variance of the historical slope sequence. When the sequence variance is greater than a preset threshold (indicating that the system is in an unstable state), the sampling frequency is automatically increased, for example, shortened to once every 5 minutes; when the sequence variance is less than the threshold (indicating that the system is in a stable state), the sampling frequency is decreased, for example, extended to once every 30 minutes.

[0103] The following features are extracted and calculated for the above time-series slope sequence K:

[0104] The sum of all slope values ​​greater than 0 in the sequence is calculated and denoted as the positive intensity index M1. This value quantifies the cumulative intensity and total momentum of the adverse trend of prolonged waiting time in high-urgency areas over a past period.

[0105] The sum of all slope values ​​less than 0 in the sequence is denoted as the negative recovery exponent M2. This value quantifies the system's cumulative ability to self-repair and optimize resource allocation, as well as its positive trend.

[0106] The number of periods in a statistical sequence with a slope value equal to 0 or an absolute value close to 0 is denoted as the stability index M3. This value reflects the persistence and frequency of the system being in a stable and healthy state.

[0107] The average length of consecutively unchanged slope signs in a sequence is denoted as the trend persistence index M4. For example, it can be used to calculate the average length of consecutive positive and negative slope segments. This value determines the inertia of a trend once it is formed; strong persistence means that more decisive intervention is needed.

[0108] S235. Input the calculated feature vector M = (positive strength index M1, negative recovery index M2, stability index M3, trend persistence index M4) into a predefined fuzzy logic rule set or a trained lightweight classifier (such as a decision tree or support vector machine SVM); output a clear system state value, ranging from 0 to 1, and map the value to the corresponding operating state, as follows: a system state value greater than 0.75 indicates an alarm state of increased imbalance; a system state value greater than 0.5 and less than or equal to 0.75 indicates a stable operating state; a system state value greater than 0.25 and less than or equal to 0.5 indicates a state of repair and optimization; a system state value less than or equal to 0.25 indicates an oscillating disorder state.

[0109] In this embodiment, a comprehensive prediction model integrating energy consumption quantification, regional risk assessment, and time cost analysis is constructed to provide multi-dimensional and high-precision decision input for route planning. First, the traditional simple mileage energy consumption estimation is abandoned, and a vehicle dynamics model is introduced to accurately calculate energy costs based on real-time load, road gradient, and speed. Second, through a regional gridded model, historical data and order waiting time are integrated to dynamically assess the urgency of nodes, thereby quantifying service delay risk into regional urgency costs. Finally, by analyzing the temporal slope trend of regional urgency and waiting time, and using an intelligent classifier to diagnose the global operating status of the system, the above scheme achieves a leap from rough estimation to precise calculation of energy costs, completes the transformation of service risk from fuzzy judgment to forward-looking quantification, and endows the system with the ability to understand the macro situation, laying a solid data foundation for subsequent adaptive optimization decisions and driving the entire recycling and scheduling process towards intelligence and precision.

[0110] S3. Based on multi-dimensional cost and system status information, an adaptive intelligent decision-making mechanism dynamically generates the globally optimal recycling path and transforms the planning results into executable instructions to drive vehicles in the physical world to complete the recycling operation, ultimately achieving efficient, low-cost, and rapid resource recycling scheduling;

[0111] S3 includes the following steps:

[0112] S31. Based on the output system state value and the corresponding state, the state identifier is converted into a continuous system state quantization value through a pre-set state policy mapping table within the system;

[0113] The system state quantization value is set to J. The system calls a preset mapping table to perform the conversion. An example mapping relationship is as follows: when the system state is in the state of increased imbalance alarm, the system state quantization value J = 2; when the system state is in the state of repair and optimization, the system state quantization value J = 0.5; when the system state is in the state of oscillation and disorder, the system state quantization value J = 0; when the system state is in the state of stable operation, the system state quantization value J = -1.

[0114] Subsequently, the system calls the preset strategy bias coefficient Vn, where n identifies the index of the cost factor, and each Vn is the strategy bias coefficient corresponding to the nth cost factor. The strategy bias coefficient is a non-negative real number and can be set through historical optimization or operational strategies, with a preferred value range of 0.5 to 3.0. The strategy bias coefficient reflects the operator's preference for the importance of different cost items under a given system state. When the system state quantification value J=0, the system automatically triggers the historical optimal weight call mechanism; when J=-1, the strategy bias coefficient Vn still uses the preferred value range of 0.5 to 3.0.

[0115] Based on the system state quantization value J and the policy bias coefficient Vn, the response value Rn of the nth cost is calculated according to the following steps. , where exp(·) represents the natural exponential function, that is, an exponential function with the mathematical constant e as the base, used to ensure that the response has a continuous and monotonically amplified or reduced effect on state changes; the exponent with Vn as the coefficient includes a bias coefficient, which can adjust the sensitivity of each factor to the quantized value J of the system state;

[0116] To ensure that the obtained weights satisfy the probability distribution characteristics, the system performs normalization on each response value Rn, making the sum of all weights equal to 1. The instantaneous weights obtained after normalization are used to reflect the theoretical weight distribution under the current control state. To avoid scheduling oscillations in short-period fluctuation scenarios, the system smooths the instantaneous weights based on exponentially weighted moving average (EWMA) to generate stable output weights. This smoothing process can effectively suppress the interference of short-term abnormal data on the weight model and improve the robustness of path calculation.

[0117] When multiple cost factors increase significantly simultaneously, the system implements a conflict handling mechanism to prevent weights from concentrating on a certain type of factor, thus causing scheduling bias imbalance. Conflict handling includes priority correction, scaling, and renormalization steps. Priority correction is used to increase the weights of key factors according to preset operational strategies. If the weight of a certain type of factor or factor group exceeds a set upper limit (e.g., 0.90) after correction, the system reduces its proportion by scaling and redistributes the remaining weights to lower priority factors. Finally, the corrected weights are normalized and smoothed to ensure that the output weights are continuous, controllable, and directly usable in the scheduling objective function.

[0118] The various parameters in this step (such as the system state quantification value J and the strategy bias coefficient Vn) can be obtained through historical data backtracking, expert experience setting, or online learning mechanisms, and can be adjusted by the operator through the management interface during system operation. This mechanism ensures the reproducibility, configurability, and stability of weight calculation, thereby making path planning adaptive and engineering feasible under different system states.

[0119] The system state quantization value and the system's preset policy bias coefficient Vi are input into the adaptive weight calculation model to dynamically generate the weights of each cost item. The weight calculation model is as follows:

[0120] Where Wn is the dynamic weight of the nth cost item, for example, n=1 corresponds to the weight of regional urgency cost, n=2 corresponds to the weight of energy cost, and n=3 corresponds to the weight of time cost; Vi is the strategy bias coefficient of the ith cost item, which is a fixed preset value and reflects the core scheduling strategy; where m in Vm is a loop variable, representing that when calculating the denominator of the formula, all weight categories need to be traversed for calculation and summation, and Vi and Vm are the same set of strategy bias coefficients;

[0121] The strategy bias coefficient Vi is a set of preset fixed parameters, the value of which is determined according to the strategy objective of the recyclable resource recovery method under different system states;

[0122] S32. Construct an objective function for path optimization, namely the comprehensive cost function, specifically:

[0123] Where C1 is the calculated fuel energy cost, or C2 is the electric energy cost, or C3 is the hybrid energy cost, with different energy costs selected depending on the vehicle type;

[0124] S33. The constructed comprehensive cost function, along with all constraints such as vehicle location, load, and inventory at recycling points, is input into a path optimization algorithm (such as a genetic algorithm or a tabu search algorithm) for calculation.

[0125] The optimization process must meet the following core constraints:

[0126] Vehicle load constraint: The sum of the recovery amounts at all nodes along the path must not exceed the vehicle's rated load capacity;

[0127] Task uniqueness constraint: Each task can be completed by at most one vehicle;

[0128] Vehicle capacity constraint: The actual load of a vehicle at any point on the route must not exceed its capacity;

[0129] Time window constraint (if applicable): The task must be completed within the specified time frame;

[0130] Output: After calculation, the algorithm outputs a set of optimal collection routes for the entire fleet. This route sequence ensures the lowest overall cost under the current system state. The termination condition is when the number of iterations reaches 200.

[0131] S34: Task dispatch and instruction issuance;

[0132] The system decomposes the optimal path sequence generated by S33 into specific, executable navigation instructions and task assignments. These instructions and tasks are then sent to the onboard terminal of the corresponding recycling vehicle or the driver's mobile application via a wireless communication network to guide the recycling operation.

[0133] In this embodiment, the system state value is quantified into a specific numerical value and input into a calculation model based on the natural exponential function and normalization processing along with a preset strategy bias coefficient. This dynamically generates real-time weights for regional urgency, energy, and time costs. Subsequently, the system constructs a comprehensive cost function based on these dynamic weights and uses a path optimization algorithm to solve for the globally optimal collection and transportation path sequence under realistic constraints such as load capacity and task uniqueness. Finally, the system automatically decomposes the optimal path into specific navigation and operation instructions and sends them to the vehicle terminal for execution. Through this step, the scheduling strategy is upgraded from static and fixed to dynamic and adaptive, which can intelligently adjust the optimization target according to the real-time status of the system. This ensures that the path planning result has both global optimality and practical feasibility, and completes precise closed-loop control from digital decision-making to physical execution. Ultimately, this significantly improves the overall efficiency, economy, and response speed of the recycling operation.

[0134] This invention addresses the core pain points of traditional recycling scheduling, such as inaccurate cost estimation, rigid strategies, and unrealistic decision-making. It constructs an integrated intelligent solution encompassing perception, decision-making, and execution. Its core concept lies in: first, accurately quantifying the three major costs—energy, time, and service urgency—through vehicle dynamics models and regional grid analysis; second, an adaptive weighted decision-making mechanism dynamically adjusts cost weights based on real-time system status (such as service imbalance alarms), enabling intelligent switching of scheduling strategies; and finally, generating a globally optimal path and automatically issuing it for execution while strictly meeting actual operational constraints. This invention represents a leap from static mechanical planning to dynamic intelligent control, significantly improving the economy, efficiency, and service quality of recycling operations.

[0135] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

Claims

1. A big data-based renewable resource recycling system, characterized in that, include; The data acquisition module obtains static basic information of the recycling network and vehicle operating parameters through the system configuration database, IoT sensing devices deployed at recycling nodes and vehicles, user terminal interfaces, and third-party map and energy service APIs; it also performs standardized preprocessing on the collected raw data and finally aggregates the multi-source data into a structured fusion dataset according to node number and time index. The resource recycling analysis module calculates the estimated total energy consumption based on vehicle dynamics models and real-time parameters, and calculates energy costs by combining energy unit prices. Simultaneously, it calculates node urgency based on gridded areas, historical data, and order waiting times, and defines regional urgency accordingly. It also estimates the total task time based on real-time traffic and node workload, and calculates the final time cost by combining unit time cost coefficients. Finally, it extracts feature vectors by analyzing the time-series slope sequence fitted by regional urgency and order waiting times, and outputs system state values ​​and specific states using a pre-built classifier. The planning and scheduling module quantifies the received system status values ​​into specific values ​​and inputs them, along with preset strategy bias coefficients, into an adaptive weight calculation model based on exponential functions and normalization processing to dynamically generate the weights of various costs. Subsequently, an optimization function with the goal of minimizing the weighted comprehensive cost is constructed, and the solution is obtained based on the path optimization algorithm under the conditions of satisfying vehicle load, task uniqueness and capacity constraints, and the optimal collection route sequence of the fleet is output. Finally, the route sequence is decomposed into navigation and operation instructions, which are sent to the vehicle terminal through the wireless communication network. The adaptive weight calculation model described above is as follows: , where Wn is the dynamic weight of the nth cost, n=1 corresponds to the weight of the regional urgency cost, n=2 corresponds to the weight of the energy cost, and n=3 corresponds to the weight of the time cost; Vi is the policy bias coefficient of the i-th cost, which is a fixed preset value and reflects the core scheduling policy; where m in Vm is a loop variable, which means that when calculating the denominator of the formula, all weight categories need to be traversed for calculation and summation. Vi and Vm are the same set of policy bias coefficients. The strategy bias coefficient Vi is a set of preset fixed parameters, the value of which is determined according to the strategy objective of the recycling method for renewable resources under different system states. In the above formula, J is the system state quantization value. Based on the output system state value and the corresponding state, the system uses a pre-set state policy mapping table to convert the state identifier into a continuous system state quantization value J.

2. The big data-based renewable resource recycling system according to claim 1, characterized in that, The static basic information and vehicle operating parameters specifically include: Static basic information includes unique identifiers for all resource recycling nodes, GPS latitude and longitude coordinates and the theoretical maximum capacity of the recycling nodes, unique identifiers for all resource recycling vehicles, the energy type of the vehicles, and the reference rolling resistance coefficient and air resistance coefficient of the vehicles. Vehicle operating parameters include real-time vehicle load capacity, rated load capacity, empty vehicle weight, equipment full load status indicator, real-time location, driving speed, remaining load capacity, vehicle battery power or fuel balance uploaded by the resource recycling node.

3. The big data-based renewable resource recycling system according to claim 1, characterized in that, The estimated total energy consumption is as follows: Based on the dynamic parameters consisting of real-time vehicle load, road gradient, and driving speed, the energy consumption of the vehicle on the road segment is estimated through an energy consumption calculation model. The energy consumption of all road segments is then summed to obtain the estimated total energy consumption to complete the current task.

4. The big data-based recycling system for renewable resources according to claim 1, characterized in that, The urgency of the area is specifically as follows: The recycling service area is divided into multiple regular grids on the electronic map. For each grid with unprocessed orders, the node urgency of the grid is calculated. The node urgency of all grids with unprocessed orders that are not covered by the candidate path is obtained. The sum of the node urgency of all such grids in the area is the area urgency.

5. A big data-based renewable resource recycling system according to claim 1, characterized in that, The final time cost is specifically as follows: The candidate path is divided into several road segments. For each road segment, the real-time average driving speed and the length of the road segment are obtained, and the estimated driving time on the road segment is calculated. For each recycling node that needs to stop on the path, the operation time at that node is estimated based on the real-time inventory weight of that node and a preset unit weight operation efficiency. The total time is the sum of the travel time on the road segment and the operation time at the node; Based on the set unit time cost coefficient, the total time is multiplied by the unit time cost coefficient to obtain the final time cost of executing the candidate path.

6. The big data-based recycling system for renewable resources according to claim 1, characterized in that, The system status values ​​are specifically as follows: Using the regional urgency value of each network as the X-axis and the average waiting time of all unprocessed orders within that network as the Y-axis, a linear regression was performed to obtain the slope of the trend line. The system performs the above fitting repeatedly with a fixed time window period to obtain a time-series slope sequence; Feature extraction is performed on the time-series slope sequence to obtain a feature vector containing a positive intensity index, a negative recovery index, a stability index, and a trend persistence index; The feature vector is input into a preset classifier, which outputs the system state value.

7. A big data-based renewable resource recycling system according to claim 1, characterized in that, The optimal collection and transportation route sequence is specifically as follows: The comprehensive cost function is calculated by weighting the regional urgency cost, energy cost, and time cost. The solution is obtained using a metaheuristic algorithm under the conditions of satisfying vehicle load constraints and task uniqueness constraints, and the optimal collection and transportation route sequence is output.

8. A big data-based renewable resource recycling system according to claim 7, characterized in that, The optimal collection and transportation route sequence is decomposed into navigation and operational instructions, specifically: The optimal collection route sequence is parsed into a series of continuous navigation segments and docking point operation instructions based on the order of nodes along the route and the corresponding operation tasks of the nodes. The navigation segment and docking point operation instructions are packaged and distributed to the corresponding vehicle terminal or the driver's mobile application via wireless communication network.

9. A method for recycling renewable resources based on big data, characterized in that, The system based on any one of claims 1-8 includes the following steps: S1. Through various sensing devices and data interfaces deployed in the recycling network, real-time static basic information and vehicle operating parameters related to the recycling of renewable resources are obtained; and standardized preprocessing is performed on the collected raw data, and finally the multi-source data is aggregated into a structured fusion dataset according to the node number and time index. S2. Based on the vehicle dynamics model and real-time parameters, the total energy consumption is estimated, and the energy cost is calculated in combination with the energy unit price. At the same time, based on the gridded area, historical data and order waiting time, the node urgency is calculated, and the area urgency cost is defined accordingly. The total task time is estimated based on real-time traffic and node workload, and the time cost is calculated in combination with the unit time cost coefficient. Finally, by analyzing the time-series slope sequence fitted by the area urgency and order waiting time, the feature vector is extracted, and the system state value and specific state are output using a pre-set classifier. S3 is used to receive system status identifiers, quantify them into specific values, and input them along with preset strategy bias coefficients into an adaptive weight calculation model based on exponential functions and normalization processing to dynamically generate the weights of various costs. Subsequently, an optimization function with the goal of minimizing the weighted comprehensive cost is constructed, and the solution is performed based on the path optimization algorithm under the conditions of satisfying vehicle load, task uniqueness, and capacity constraints, outputting the optimal collection route sequence of the fleet. Finally, the optimal collection route sequence is decomposed into navigation and operation instructions, which are sent to the vehicle terminals through the wireless communication network.

Images