Logistics vehicle bill cost prediction method and device, equipment and storage medium

Abstract

The invention relates to the technical field of bill cost analysis, and discloses a logistics vehicle bill cost prediction method and device, equipment and a storage medium. The method comprises the following steps: acquiring driving data and fee data of a logistics vehicle; preprocessing the driving data and the cost data by adopting a big data processing technology, and uploading the data to a block chain; learning the processed driving data and cost data based on a trained data enhancement model to obtain an optimal action strategy; and utilizing the trained bill cost prediction model to predict the logistics vehicle bill cost of the optimal action strategy. According to the method, by means of a neural network deep learning and reinforcement learning algorithm, an expense change rule can be mined from a large amount of historical driving data and expense data, more accurate logistics vehicle bill expense prediction is made according to the expense change rule, an enterprise is helped to make budget planning in advance, and the uncertainty risk is reduced; based on the prediction result, the system can also automatically adjust the scheduling plan of the logistics vehicle.

Description

Technical Field

[0001] The present invention relates to the technical field of bill expense analysis, and in particular to a method, device, equipment and storage medium for predicting bill expenses of a logistics vehicle. Background Art

[0002] In the cost management of transportation, traditional bill cost analysis mainly relies on manual review, which has many shortcomings, such as low efficiency, error-prone, and failure to detect potential problems in a timely manner. In addition, due to the lack of effective data analysis tools, companies find it difficult to respond quickly to emergencies, resulting in additional costs. Although there are some bill analysis software on the market, most of them are limited to basic statistical functions and cannot provide in-depth cost forecasts and optimization suggestions. In addition, due to data security issues, many companies have concerns about using third-party platforms and are worried about the leakage of sensitive information. Therefore, there is an urgent need for a new method of logistics vehicle bill cost analysis to overcome the problems of existing technologies and improve the scientificity and effectiveness of cost management.

[0003] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0004] The main purpose of the present invention is to solve the problem that the existing bill analysis method is limited to basic statistical functions and cannot provide in-depth cost forecasting and optimization suggestions.

[0005] A first aspect of the present invention provides a method for predicting bill costs of logistics vehicles, comprising: obtaining driving data and cost data of logistics vehicles, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the logistics vehicle's freight bill; pre-processing the driving data and cost data using big data processing technology, and uploading them to a blockchain; learning the processed driving data and cost data based on a trained data enhancement model to learn the best action strategy; and using the trained bill cost prediction model to predict the bill costs of the logistics vehicle with the best action strategy.

[0006] Optionally, in a first implementation method of the first aspect of the present invention, the driving data and cost data of the logistics vehicle are obtained, including the steps of: the logistics vehicle is pre-equipped with a GPS locator, a fuel consumption sensor, and a load sensor, the average driving speed, driving distance and driving route of the vehicle are obtained through the GPS locator, the fuel consumption is monitored in real time through the fuel consumption sensor, and the vehicle load is obtained through the load sensor; weather data, traffic condition data and real-time oil price data are retrieved through an Internet API; the shuttle fees, freight bill quality control fines, fuel costs, tolls and parking fees of the logistics vehicle are obtained according to the freight bill of the logistics vehicle, and the overall cost of the freight bill is calculated.

[0007] Optionally, in a second implementation method of the first aspect of the present invention, the driving data and the fee data are pre-processed using big data processing technology and then uploaded to the blockchain, including the steps of: cleaning, converting and loading the driving data and the fee data through an ETL tool to finally form unified pre-processed data; converting the pre-processed data into a format suitable for storage on the blockchain; writing a smart contract to define data storage and operation rules; deploying the smart contract to the blockchain platform, and calling the method in the smart contract to upload the pre-processed data to the blockchain.

[0008] Optionally, in a third implementation method of the first aspect of the present invention, the processed logistics vehicle data is learned based on a trained data enhancement model to learn the best action strategy, including the steps of: collecting historical driving data and cost data of the logistics vehicle, wherein the historical driving data includes vehicle type, average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data includes the overall cost of the logistics vehicle's freight bill; standardizing the historical driving data and cost-related data to ensure that data of different dimensions are comparable; constructing a data enhancement learning model based on a reinforcement learning algorithm; training the data enhancement learning model according to the historical driving data and cost-related data of the logistics vehicle, allowing the data enhancement learning model to learn to take corresponding actions under different states to obtain the maximum reward, thereby obtaining a trained data enhancement model; inputting the driving data of the current logistics vehicle into the trained data enhancement model for learning, and outputting the best action strategy for the vehicle under different states.

[0009] Optionally, in a fourth implementation method of the first aspect of the present invention, a data enhancement learning model is constructed based on a reinforcement learning algorithm, including the steps of: constructing a data enhancement learning model based on a Q-Learning reinforcement learning algorithm, wherein the update formula of Q-Learning is: Q(s,a)←Q(s,a)+α[r+γa′maxQ(s′,a′)-Q(s,a)], wherein s and s′ respectively represent the current state and the next state, a and a′ respectively represent the current action and the action that may be taken in the next state, r is the immediate reward obtained after executing action a, α is the learning rate, and γ is the discount factor; defining the state, action and reward of the logistics vehicle in the data enhancement model, wherein defining the state of the logistics vehicle includes: the starting position of the vehicle, the terminal position of the vehicle, the type of vehicle, the load of the vehicle, and the fuel consumption of the vehicle; defining the action of the logistics vehicle includes: the driving route and the average driving speed; defining the reward obtained after the vehicle takes a certain action as: the reduced overall cost of the freight bill.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, a trained neural network model is used to predict the bill expense trend of a logistics vehicle for an optimal action strategy, including the steps of: collecting historical expense data of logistics vehicles under different action strategies; constructing an initial bill expense prediction model based on a convolutional neural network; using historical expense data as labels and different action strategies of logistics vehicles as feature data to train the initial bill expense prediction model, calculating predicted values ​​through forward propagation, adjusting weights through back propagation, minimizing the error between predicted values ​​and actual values, and thereby obtaining a trained bill expense prediction model; inputting the optimal action strategy of the logistics vehicle into the trained bill expense prediction model, and outputting predicted bill expenses of the logistics vehicle.

[0011] Optionally, in the sixth implementation method of the first aspect of the present invention, the steps are also included: predicting the total bill costs of the logistics vehicles within the next month under the optimal action strategy conditions based on the trained bill cost prediction model; if the predicted total bill costs of the logistics vehicles within the next month exceeds the total budget, readjusting the itinerary of the logistics vehicles.

[0012] The second aspect of the present invention provides a bill cost prediction device for a logistics vehicle, comprising: an acquisition module, used to acquire driving data and cost data of the logistics vehicle, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the logistics vehicle's freight bill; a preprocessing module, used to preprocess the driving data and cost data using big data processing technology, and then upload them to the blockchain; a learning module, used to learn the processed driving data and cost data based on a trained data enhancement model to learn the best action strategy; a prediction module, used to use the trained bill cost prediction model to predict the bill cost of the logistics vehicle with the best action strategy.

[0013] Optionally, in a first implementation method of the second aspect of the present invention, the acquisition module includes: a first acquisition unit, used to pre-equip the logistics vehicle with a GPS locator, a fuel consumption sensor, and a load sensor, to obtain the vehicle's average driving speed, driving distance and driving route through the GPS locator, to monitor fuel consumption in real time through the fuel consumption sensor, and to obtain the vehicle's load through the load sensor; a second acquisition unit, used to retrieve weather data, traffic conditions data and real-time oil price data through an Internet API; a third acquisition unit, used to obtain the logistics vehicle's shuttle fees, freight bill quality control fines, fuel costs, tolls and parking fees based on the logistics vehicle's freight bill, and calculate the overall freight bill cost.

[0014] Optionally, in a second implementation of the second aspect of the present invention, the preprocessing module includes: a cleaning unit, used to clean, convert and load the driving data and fee data through an ETL tool, and finally form unified preprocessed data; a conversion unit, used to convert the preprocessed data into a format suitable for storage on a blockchain; a writing unit, used to write a smart contract to define data storage and operation rules; an uploading unit, used to deploy a smart contract to a blockchain platform, and call methods in the smart contract to upload the preprocessed data to the blockchain.

[0015] Optionally, in a third implementation of the second aspect of the present invention, the learning module includes: a first collection unit, used to collect historical driving data and cost data of logistics vehicles, wherein the historical driving data includes vehicle type, average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data includes the overall cost of the freight bill of the logistics vehicle; a standardization processing unit, used to standardize the historical driving data and cost-related data to ensure that data of different dimensions are comparable; a learning model construction unit, used to construct a data enhancement learning model based on a reinforcement learning algorithm; a learning model training unit, used to train the data enhancement learning model according to the historical driving data and cost-related data of the logistics vehicle, so that the data enhancement learning model learns to take corresponding actions under different states to obtain the maximum reward, thereby obtaining a trained data enhancement model; an output unit, used to input the driving data of the current logistics vehicle into the trained data enhancement model for learning, and output the best action strategy for the vehicle under different states.

[0016] Optionally, in a fourth implementation manner of the second aspect of the present invention, the learning model construction unit includes: an algorithm selection subunit, used to construct a data enhancement learning model based on the Q-Learning enhancement learning algorithm, wherein the update formula of Q-Learning is: Q(s,a)←Q(s,a)+α[r+γa′max Q(s′,a′)-Q(s,a)], wherein s and s′ respectively represent the current state and the next state, a and a′ respectively represent the current action and the action that may be taken in the next state, r is the immediate reward obtained after executing action a, α is the learning rate, and γ is the discount factor; a definition subunit, used to define the state, action and reward of the logistics vehicle in the data enhancement model, wherein the state of the logistics vehicle is defined to include: the starting position of the vehicle, the terminal position of the vehicle, the type of vehicle, the load of the vehicle, and the fuel consumption of the vehicle; the action of the logistics vehicle is defined to include: the driving route and the average driving speed; the reward obtained after the vehicle takes a certain action is defined to be: the reduced overall cost of the freight bill.

[0017] Optionally, in a fifth implementation of the second aspect of the present invention, the prediction module includes: a second collection unit, used to collect historical cost data of logistics vehicles under different action strategies; a prediction model construction unit, used to construct an initial bill cost prediction model based on a convolutional neural network; a prediction model training unit, used to use historical cost data as labels and different action strategies of logistics vehicles as feature data to train the initial bill cost prediction model, calculate the predicted value through forward propagation, adjust the weight through back propagation, minimize the error between the predicted value and the actual value, and thus obtain a trained bill cost prediction model; a prediction unit, used to input the optimal action strategy of the logistics vehicle into the trained bill cost prediction model, and output the predicted bill cost of the logistics vehicle.

[0018] Optionally, in a sixth implementation method of the second aspect of the present invention, it also includes: an adjustment module, used to predict the total bill costs of logistics vehicles within the next month under the optimal action strategy conditions based on the trained bill cost prediction model; if the predicted total bill costs of logistics vehicles within the next month exceeds the total budget, the itinerary of the logistics vehicle is readjusted.

[0019] A third aspect of the present invention provides a bill expense prediction device for a logistics vehicle, comprising: a memory and at least one processor, wherein the memory stores computer-readable instructions, and the memory and the at least one processor are interconnected via lines; the at least one processor calls the computer-readable instructions in the memory so that the bill expense prediction device for the logistics vehicle executes each step of the bill expense prediction method for the logistics vehicle as described above.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, in which computer-readable instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the various steps of the method for predicting bill costs for logistics vehicles as described above.

[0021] Beneficial effects: The technical solution of the present invention obtains the driving data and cost data of logistics vehicles, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the freight bill of the logistics vehicle; the driving data and cost data are pre-processed by big data processing technology and uploaded to the blockchain; the processed driving data and cost data are learned based on the trained data enhancement model to learn the best action strategy; the bill cost of the logistics vehicle with the best action strategy is predicted using the trained bill cost prediction model. The method of the present invention, with the help of neural network deep learning and reinforcement learning algorithms, can mine the cost change rules from a large amount of historical driving data and cost data, and make more accurate bill cost predictions for logistics vehicles based on this, which helps enterprises to make budget planning in advance and reduce uncertainty risks; based on the prediction results, the system can also automatically adjust the scheduling plan of logistics vehicles, select the optimal route, avoid congested periods, and thus reduce unnecessary expenses. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A first flow chart of a method for predicting bill costs of a logistics vehicle provided in an embodiment of the present invention;

[0023] Figure 2 A second flow chart of the method for predicting bill costs of logistics vehicles provided in an embodiment of the present invention;

[0024] Figure 3 A third flow chart of the method for predicting bill costs of logistics vehicles provided in an embodiment of the present invention;

[0025] Figure 4 A fourth flow chart of a method for predicting bill costs of a logistics vehicle provided in an embodiment of the present invention;

[0026] Figure 5 A fifth flow chart of the method for predicting bill costs of logistics vehicles provided in an embodiment of the present invention;

[0027] Figure 6 A sixth flow chart of the method for predicting bill costs of logistics vehicles provided in an embodiment of the present invention;

[0028] Figure 7 A schematic diagram of a structure of a bill expense prediction device for a logistics vehicle provided by an embodiment of the present invention;

[0029] Figure 8 Another structural schematic diagram of a bill expense prediction device for a logistics vehicle provided by an embodiment of the present invention;

[0030] Fig. 9A schematic diagram of the structure of a bill expense prediction device for a logistics vehicle provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The embodiment of the present invention provides a method, device, equipment and storage medium for predicting the billing costs of logistics vehicles, obtaining the driving data and cost data of logistics vehicles, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the freight bill of the logistics vehicle; the driving data and cost data are pre-processed by big data processing technology and uploaded to the blockchain; the processed driving data and cost data are learned based on the trained data enhancement model to learn the best action strategy; the billing costs of logistics vehicles with the best action strategy are predicted using the trained billing cost prediction model. The method of the present invention can mine the cost change rules from a large amount of historical driving data and cost data with the help of neural network deep learning and reinforcement learning algorithms, and make more accurate billing cost predictions for logistics vehicles based on this, which helps enterprises to make budget planning in advance and reduce uncertainty risks; based on the prediction results, the system can also automatically adjust the scheduling plan of logistics vehicles, select the best route, avoid congestion periods, and thus reduce unnecessary expenses.

[0032] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 The first embodiment of the method for predicting bill costs of logistics vehicles in the embodiment of the present invention includes:

[0034] S100, obtaining driving data and cost data of logistics vehicles, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the freight bill of the logistics vehicle;

[0035] Specifically, the average driving speed, vehicle load, fuel consumption, driving distance and other key factors directly related to logistics costs in driving data. The comprehensive consideration of these factors enables the cost prediction model to more comprehensively capture the root causes of cost changes, thereby improving the accuracy and reliability of the prediction; changes in external factors such as weather data, traffic conditions data and real-time oil price data also have a significant impact on logistics costs. Incorporating these factors into the prediction model can more accurately reflect the complexity of the real world and make the prediction results closer to the actual situation. Different types of driving data and cost data provide rich training samples for the model, which helps the model learn more diverse feature representations, which enables the subsequent network model to better generalize when facing new and unknown data, thereby improving the versatility and robustness of the prediction. Real-time acquisition and analysis of driving data and cost data enable enterprises to achieve refined management of the logistics process. By monitoring and optimizing each link, enterprises can improve logistics efficiency and reduce operating costs. Based on the results of cost prediction, enterprises can optimize the scheduling and route planning of logistics vehicles, reduce empty driving rate and waiting time, and improve vehicle utilization and transportation efficiency.

[0036] S200, pre-processing the driving data and the fee data using big data processing technology, and uploading them to the blockchain;

[0037] Specifically, invalid, duplicate, erroneous or abnormal data records will be removed during the preprocessing process. If these problematic data are not processed, they may interfere with the model's learning process and lead to inaccurate prediction results. Data from different sources and formats will be unified under the same standard to ensure data consistency and comparability, which will help with subsequent data analysis and model training. In the preprocessing stage, sensitive data can be encrypted to protect data privacy and security, which is especially important for data uploaded to the blockchain, because the immutability of the blockchain requires that data is difficult to change once uploaded, so it is necessary to ensure that the data is safe and reliable before uploading.

[0038] S300, learning the processed driving data and fare data based on the trained data enhancement model to obtain the best action strategy;

[0039] In this embodiment, the processed driving data and cost data are learned based on the data enhancement model, and the optimal action strategy is learned, which can not only improve the generalization ability of the model and the accuracy of prediction, but also optimize the logistics and transportation strategy, enhance the reliability of decision-making, and support real-time optimization and improve competitiveness.

[0040] S400: Use the trained bill cost prediction model to predict the logistics vehicle bill cost of the best action strategy.

[0041] In this embodiment, the prepared optimal action strategy is input into the trained bill cost prediction model, and the model predicts future bill costs based on the input data and the learned rules; based on the prediction results, the impact of different action strategies (such as choosing different driving routes, adjusting loads, optimizing speed, etc.) on bill costs can be evaluated; by comparing the predicted costs of different strategies, the optimal action strategy can be selected, that is, the strategy that can minimize the bill costs. The optimal action strategy is applied to actual logistics transportation, and the difference between the actual bill costs and the predicted values ​​is monitored; based on the monitoring results, the model can be further tuned to improve the accuracy and reliability of the prediction.

[0042] The method of the present invention, with the help of neural network deep learning and reinforcement learning algorithms, can mine the cost change patterns from a large amount of historical driving data and cost data, and make more accurate bill cost forecasts for logistics vehicles based on this, which helps companies to make budget plans in advance and reduce uncertainty risks; based on the prediction results, the system can also automatically adjust the scheduling plan of logistics vehicles, select the optimal route, avoid congested periods, and thus reduce unnecessary expenses.

[0043] See also Figure 2 The second embodiment of the method for predicting bill costs of logistics vehicles in the embodiment of the present invention includes:

[0044] S110, the logistics vehicle is pre-equipped with a GPS locator, a fuel consumption sensor, and a load sensor, and the average driving speed, driving distance, and driving route of the vehicle are obtained through the GPS locator, the fuel consumption is monitored in real time through the fuel consumption sensor, and the vehicle load is obtained through the load sensor;

[0045] S120, retrieving weather data, traffic data, and real-time oil price data through an Internet API;

[0046] S130. Obtain the shuttle fee, freight bill quality control fine, fuel cost, toll and parking fee of the logistics vehicle according to the freight bill of the logistics vehicle, and calculate the overall cost of the freight bill.

[0047] Specifically, select a high-precision, low-energy GPS locator to ensure that it can work stably in various environments, install a fuel consumption sensor, directly connected to the vehicle's fuel system, to monitor fuel consumption in real time; load weight sensors are usually installed on the vehicle's suspension system or the bottom of the cargo box to accurately measure the load, and ensure that all sensors communicate with the cloud server wirelessly (such as 4G / 5G, etc.); set the data collection frequency, such as collecting GPS location, speed, direction and other information every 5 minutes; fuel consumption and load data can be collected according to actual needs. The collection frequency can be set, such as every certain distance or time. Connect with weather service providers to obtain real-time and future weather forecasts; cooperate with transportation departments or third-party data providers to obtain traffic road condition data (such as congestion, construction information, etc.); access real-time oil price API to obtain gas station oil price information. This embodiment integrates the acquired weather, road condition, and oil price data with vehicle driving data, which is conducive to the later prediction of the best driving route, considering weather, road conditions and oil price factors, and optimizing transportation plans; it can also avoid bad weather and traffic congestion in advance, reducing transportation time and cost.

[0048] In this embodiment, fixed cost information such as shuttle fares and quality control fines can be automatically extracted from the freight bill system; fuel costs can be calculated based on GPS and fuel consumption sensor data; relevant costs can be counted by combining toll invoices and parking fee receipts (or automatic payment records); a cost calculation algorithm is developed to summarize various costs, derive the overall cost of the freight bill, and generate a detailed cost report, including cost details, cost analysis, etc. Automated cost calculation reduces manual errors and improves financial processing efficiency.

[0049] See also Figure 3 The third embodiment of the method for predicting bill costs of logistics vehicles in the embodiment of the present invention includes:

[0050] S210, cleaning, converting and loading the driving data and the cost data through an ETL tool to finally form unified pre-processed data;

[0051] S220, converting the preprocessed data into a format suitable for storage on a blockchain;

[0052] S230. Write smart contracts to define data storage and operation rules;

[0053] S240, deploy the smart contract to the blockchain platform, and call the method in the smart contract to upload the pre-processed data to the blockchain.

[0054] Specifically, first, collect driving data and cost data from the GPS locator, fuel consumption sensor, load sensor and other equipment of the logistics vehicle and the freight bill system. Use ETL tools (such as ETLCloud) to clean the collected data and remove noise data, invalid values ​​and missing values. For example, check the abnormal points in the GPS data (such as abnormally high or low speed), handle the sudden increase or decrease in fuel consumption data, etc. According to business needs, the cleaned data is format converted and standardized. For example, the dates in different time formats are uniformly converted to the ISO 8601 standard format, and the fuel consumption data is converted from the original unit to the standard unit (such as liters / hundred kilometers); the converted data is loaded into the data warehouse or database for subsequent analysis and processing. During the loading process, optimization measures such as data partitioning and indexing can be set to improve data query efficiency. This embodiment ensures the accuracy and consistency of the data through cleaning and conversion, and provides a reliable data basis for subsequent analysis; through standardized data formats, it is convenient to share and exchange data between different systems and applications, and reduce the difficulty of data integration; through reasonable data partitioning and index settings, the efficiency of data query and analysis is improved.

[0055] In this embodiment, according to the requirements of the blockchain platform, a data structure suitable for storage is defined. Generally, the data on the blockchain needs to be stored in a format that is easy to parse and verify, such as JSON and XML. The preprocessed data is converted into a format supported by the blockchain platform, which includes converting the data table into a JSON object or XML document, and ensuring that each field complies with the data type and constraint requirements of the blockchain platform. During the conversion process, the data is verified to ensure the integrity and consistency of the data. For example, check whether the key-value pairs of the JSON object are complete, whether there are duplicate fields, etc. In this embodiment, the blockchain platform usually has a powerful data verification mechanism to ensure that the data stored on the blockchain is not tampered with. The distributed ledger characteristics of the blockchain make it difficult for the data to be maliciously tampered with or deleted, which improves the security of the data. At the same time, the data on the blockchain can be accessed and verified by multiple participants, which facilitates the sharing and transparency of data.

[0056] In this embodiment, the functions and rules that the smart contract needs to implement are clarified. For example, it is necessary to define which data can be stored on the blockchain, which users have access to the data, and how the data is updated and deleted. Use smart contract programming languages ​​such as Solidity to write code, define data structures, functions, events, etc. in the contract to define the storage and operation rules of the data; fully test the smart contract before deployment to ensure that it functions correctly, has stable performance, and has no security vulnerabilities. Smart contracts can automatically execute preset rules and logic without human intervention, which improves efficiency and accuracy. Smart contracts run on the blockchain and do not rely on any centralized organization, which improves the reliability and anti-attack capabilities of the system.

[0057] In this embodiment, a suitable blockchain platform (such as Ethereum, EOS, etc.) is selected according to business needs. , The smart contract is deployed to the blockchain using the development tools of the blockchain platform (such as Remix IDE, Truffle, etc.). , The method in the smart contract is called through the API of the blockchain platform or the wallet software, and the pre-processed data is uploaded to the blockchain. During the upload process, it is necessary to ensure that the format and content of the data meet the requirements of the smart contract. In this embodiment, once the data is uploaded to the blockchain, it cannot be tampered with or deleted, which ensures the authenticity and credibility of the data; at the same time, the data on the blockchain has characteristics such as timestamps and transaction records, which facilitates the tracing and auditing of the data; the distributed ledger characteristics of the blockchain make the data more credible and reliable, which helps to establish a trust mechanism.

[0058] See also Figure 4 The fourth embodiment of the method for predicting bill costs of express logistics vehicles in the embodiment of the present invention includes:

[0059] S310, collecting historical driving data and cost data of logistics vehicles, wherein the historical driving data includes vehicle type, average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data includes the overall cost of the freight bill of the logistics vehicle;

[0060] S320, standardizing the historical driving data and cost-related data to ensure that data of different dimensions are comparable;

[0061] S330, constructing a data enhancement learning model based on the enhancement learning algorithm;

[0062] S340, training the data enhancement learning model according to the historical driving data and cost-related data of the logistics vehicle, allowing the data enhancement learning model to learn to take corresponding actions in different states to obtain the maximum reward, thereby obtaining a trained data enhancement model;

[0063] S350: Input the driving data of the current logistics vehicle into the trained data enhancement model for learning, and output the optimal action strategy for the vehicle under different conditions.

[0064] Specifically, this embodiment constructs a data enhancement learning model based on the Q-Learning enhancement learning algorithm, wherein the update formula of Q-Learning is: Q(s,a)←Q(s,a)+α[r+γa′max Q(s′,a′)-Q(s,a)], wherein s and s′ represent the current state and the next state respectively, a and a′ represent the current action and the action that may be taken in the next state respectively, r is the immediate reward obtained after executing action a, α is the learning rate, and γ is the discount factor; the state, action and reward of the logistics vehicle in the data enhancement model are defined, wherein the state of the logistics vehicle is defined to include: the starting position of the vehicle, the terminal position of the vehicle, the type of vehicle, the load of the vehicle, and the fuel consumption of the vehicle, and these state information will be used to describe the configuration of the vehicle at any given point in time; the action of the logistics vehicle is defined to include: the driving route and the average driving speed; the reward obtained after the vehicle takes a certain action is defined as: the reward function is designed according to the overall cost of the freight bill reduced after the logistics vehicle takes a certain action. For example, if an action reduces the cost by 100 yuan, a corresponding positive reward is given; if the cost increases or remains unchanged, a smaller reward or penalty is given. Through the reward function, the algorithm can clearly know which actions are beneficial and which are disadvantageous, thereby guiding the learning process to optimize in the direction of reducing costs; Initialize the Q table: Create a Q value table for each possible state-action pair and initialize it to zero or a small random number. This table will be used to store the Q values ​​during the learning process.

[0065] This embodiment uses the historical driving data and cost data of logistics vehicles to train the model. These data should contain a variety of different driving scenarios and conditions so that the model can learn a wide range of experience. Through continuous trial and error and learning, the Q-Learning algorithm can gradually find the best action strategy to take under different conditions, thereby optimizing the driving cost of logistics vehicles.

[0066] See also Figure 5 The fifth embodiment of the bill cost prediction of the logistics vehicle in the embodiment of the present invention includes:

[0067] S410, collecting historical cost data of logistics vehicles under different action strategies;

[0068] S420, constructing an initial bill cost prediction model based on a convolutional neural network;

[0069] S430, using historical cost data as labels and different action strategies of logistics vehicles as feature data, training the initial bill cost prediction model, calculating predicted values ​​through forward propagation, adjusting weights through back propagation, minimizing the error between the predicted values ​​and the actual values, thereby obtaining a trained bill cost prediction model;

[0070] S440, input the optimal action strategy of the logistics vehicle into the trained bill cost prediction model, and output the predicted bill cost of the logistics vehicle.

[0071] Specifically, the historical cost data of logistics vehicles under different action strategies are collected from the logistics management system, and duplicate, erroneous or inconsistent data records are removed; the corresponding action strategy (such as driving route, average driving speed, etc.) and corresponding costs (such as fuel costs, tolls, maintenance fees, etc.) are marked for each data record; the cleaned and labeled data are stored in the database.

[0072] Although convolutional neural networks are usually used for image processing, this embodiment applies them to one-dimensional data (such as time series or feature vectors) and adjusts the network structure to adapt to the cost prediction task. Its input layer: receives different action strategies of logistics vehicles as feature data. These features may include driving distance, average speed, route type, load, etc.; convolution layer: used to extract local correlation between features, and a one-dimensional convolution kernel can be used; pooling layer (optional): used to reduce feature dimensions, reduce calculation amount, and prevent overfitting; fully connected layer: flattens the output of the convolution layer and pooling layer, and performs nonlinear transformation through the fully connected layer to learn the complex relationship between features; output layer: outputs the predicted logistics vehicle bill. Among them, the convolution layer outputs: y i =σ(∑ j w j x i+j +b), where x is the input feature, w is the convolution kernel weight, b is the bias term, σ is the activation function (such as ReLU; fully connected layer output: z = σ(W y + b′), where y is the output of the convolution layer (or pooling layer), W is the weight matrix, b′ is the bias term, and σ is the activation function.

[0073] Divide the historical cost data under different action strategies into training set, validation set and test set; use the training set data to train the model and calculate the predicted value through forward propagation; use the loss function (such as mean square error MSE) to calculate the error between the predicted value and the actual value, and adjust the model weight through the back propagation algorithm to minimize the error; repeat the above process until the preset training round or error threshold is reached. Use the validation set data to evaluate the performance of the model to ensure that the model is not overfitting. The training process enables the model to learn the rules in the historical data and improve the prediction accuracy. The validation process ensures the generalization ability of the model and prevents the occurrence of overfitting.

[0074] The optimal action strategy of the logistics vehicle is input into the trained bill cost prediction model, and the model outputs the predicted logistics vehicle bill cost based on the input optimal action strategy. The prediction results can help logistics companies plan budgets in advance and optimize cost structures; by comparing the predicted costs under different action strategies, logistics companies can choose the strategy with the lowest cost to execute.

[0075] See also Figure 6 The sixth embodiment of the bill cost prediction of the logistics vehicle in the embodiment of the present invention includes:

[0076] S510, predicting the total bill cost of the logistics vehicle within the next month under the optimal action strategy condition according to the trained bill cost prediction model;

[0077] S520. If the total bill costs of the logistics vehicles predicted within the next month exceed the total budget, the itinerary of the logistics vehicles is readjusted.

[0078] Specifically, the prepared input data is input into the trained bill cost prediction model. Based on these input data and the rules it has learned, the model will predict the logistics vehicle bill costs for each day of the next month; the predicted costs for each day will be summarized to obtain the total logistics vehicle bill costs for the next month.

[0079] Compare the predicted total bill costs of logistics vehicles within the next month with the preset total budget. If the predicted costs exceed the total budget, the logistics vehicle schedule needs to be readjusted. For example, try to optimize the driving route, choose a lower-cost route or adjust the driving time to avoid peak hours; consider reducing unnecessary loads or adjusting the vehicle type to reduce costs; regenerate a new schedule and use the bill cost prediction model again to make predictions to ensure that the costs under the new schedule are within the budget; after confirming that the new schedule can meet business needs and the costs are within the budget, execute the new schedule.

[0080] Through timely adjustments, logistics companies can avoid financial risks caused by cost overruns; readjustment of itineraries can further optimize cost structures and improve logistics efficiency; this approach helps logistics companies maintain flexibility and adaptability in the face of uncertainty.

[0081] The above describes the method for predicting the billing fee of a logistics vehicle in an embodiment of the present invention. The following describes the device for predicting the billing fee of a logistics vehicle in an embodiment of the present invention. Figure 7 In one embodiment of the present invention, a bill expense prediction device for a logistics vehicle includes:

[0082] The acquisition module 10 is used to acquire the driving data and cost data of the logistics vehicle, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the freight bill of the logistics vehicle;

[0083] A preprocessing module 20, used to preprocess the driving data and the cost data using big data processing technology and upload them to the blockchain;

[0084] A learning module 30, for learning the processed driving data and fare data based on the trained data enhancement model to obtain the best action strategy;

[0085] The prediction module 40 is used to predict the bill cost of the logistics vehicle for the best action strategy by using the trained bill cost prediction model.

[0086] The method of the present invention, with the help of neural network deep learning and reinforcement learning algorithms, can mine the cost change patterns from a large amount of historical driving data and cost data, and make more accurate bill cost predictions for logistics vehicles based on this, which helps companies to make budget plans in advance and reduce uncertainty risks; based on the prediction results, the system can also automatically adjust the scheduling plan of logistics vehicles, select the optimal route, avoid congested periods, and thus reduce unnecessary expenses.

[0087] See also Figure 8 Another embodiment of the bill expense prediction device for a logistics vehicle in the embodiment of the present invention includes:

[0088] The acquisition module 10 is used to acquire the driving data and cost data of the logistics vehicle, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the freight bill of the logistics vehicle;

[0089] A preprocessing module 20, used to preprocess the driving data and the cost data using big data processing technology and upload them to the blockchain;

[0090] A learning module 30, for learning the processed driving data and fare data based on the trained data enhancement model to obtain the best action strategy;

[0091] The prediction module 40 is used to predict the bill cost of the logistics vehicle for the best action strategy by using the trained bill cost prediction model.

[0092] In this embodiment, the acquisition module 10 includes:

[0093] The first acquisition unit 11 is used for pre-equipping the logistics vehicle with a GPS locator, a fuel consumption sensor, and a load sensor, obtaining the average driving speed, driving distance, and driving route of the vehicle through the GPS locator, monitoring the fuel consumption in real time through the fuel consumption sensor, and obtaining the vehicle load through the load sensor;

[0094] The second acquisition unit 12 is used to retrieve weather data, traffic data and real-time oil price data through an Internet API;

[0095] The third acquisition unit 13 is used to obtain the shuttle fee, freight bill quality control fine, fuel cost, toll and parking fee of the logistics vehicle according to the freight bill of the logistics vehicle, and calculate the overall cost of the freight bill.

[0096] In this embodiment, the preprocessing module 20 includes:

[0097] A cleaning unit 21, used for cleaning, converting and loading the driving data and the cost data through an ETL tool, and finally forming unified pre-processed data;

[0098] A conversion unit 22, used to convert the pre-processed data into a format suitable for storage on the blockchain; a writing unit, used to write a smart contract to define the storage and operation rules of the data;

[0099] The uploading unit 23 is used to deploy the smart contract to the blockchain platform and call the method in the smart contract to upload the pre-processed data to the blockchain.

[0100] In this embodiment, the learning module 30 includes:

[0101] The first collection unit 31 is used to collect historical driving data and cost data of logistics vehicles, wherein the historical driving data includes vehicle type, average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data includes the overall cost of the freight bill of the logistics vehicle;

[0102] A standardization processing unit 32 is used to perform standardization processing on the historical driving data and the cost-related data to ensure that the data of different dimensions are comparable;

[0103] A learning model building unit 33, used to build a data enhancement learning model based on an enhancement learning algorithm;

[0104] The learning model training unit 34 is used to train the data enhancement learning model according to the historical driving data and cost-related data of the logistics vehicle, so that the data enhancement learning model learns to take corresponding actions in different states to obtain the maximum reward, thereby obtaining a trained data enhancement model;

[0105] The output unit 35 is used to input the driving data of the current logistics vehicle into the trained data enhancement model for learning, and output the optimal action strategy to be taken by the vehicle under different conditions.

[0106] In this embodiment, the learning model building unit 33 includes:

[0107] The algorithm selection subunit 331 is used to construct a data enhancement learning model based on the Q-Learning enhancement learning algorithm, wherein the update formula of Q-Learning is: Q(s,a)←Q(s,a)+α[r+γa′max Q(s′,a′)-Q(s,a)], wherein s and s′ represent the current state and the next state respectively, a and a′ represent the current action and the possible actions to be taken in the next state respectively, r is the immediate reward obtained after executing action a, α is the learning rate, and γ is the discount factor;

[0108] The definition subunit 332 is used to define the state, action and reward of the logistics vehicle in the data enhancement model, wherein the state of the logistics vehicle is defined to include: the starting position of the vehicle, the terminal position of the vehicle, the type of vehicle, the vehicle load, and the fuel consumption of the vehicle; the action of the logistics vehicle is defined to include: the driving route and the average driving speed; and the reward obtained by the vehicle after taking a certain action is defined to be: the overall cost of the freight bill is reduced.

[0109] In this embodiment, the prediction module 40 includes:

[0110] The second collection unit 41 is used to collect historical cost data of logistics vehicles under different action strategies;

[0111] A prediction model building unit 42, used to build an initial bill cost prediction model based on a convolutional neural network;

[0112] The prediction model training unit 43 is used to train the initial bill cost prediction model using the historical cost data as labels and the different action strategies of the logistics vehicles as feature data, calculate the prediction value through forward propagation, adjust the weight through back propagation, minimize the error between the prediction value and the actual value, and thus obtain the trained bill cost prediction model;

[0113] The prediction unit 44 is used to input the optimal action strategy of the logistics vehicle into the trained bill cost prediction model and output the predicted logistics vehicle bill cost.

[0114] In this embodiment, it also includes: an adjustment module 50, which is used to predict the total bill costs of the logistics vehicles within the next month under the optimal action strategy conditions based on the trained bill cost prediction model; if the predicted total bill costs of the logistics vehicles within the next month exceeds the total budget, the itinerary of the logistics vehicles is readjusted.

[0115] above Figure 7 and Figure 8 The bill expense prediction device for the logistics vehicle in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The bill expense prediction device for the logistics vehicle in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0116] Fig. 9 It is a structural diagram of a bill expense prediction device for a logistics vehicle provided by an embodiment of the present invention. The bill expense prediction device 100 for a logistics vehicle may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 11 (for example, one or more processors) and a memory 12, and one or more storage media 13 (for example, one or more massive storage devices) storing application programs 133 or data 132. Among them, the memory 12 and the storage medium 13 can be short-term storage or permanent storage. The program stored in the storage medium 13 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the bill expense prediction device 100 for a logistics vehicle. Furthermore, the processor 11 can be configured to communicate with the storage medium 13 to execute a series of instruction operations in the storage medium 13 on the bill expense prediction device 100 for a logistics vehicle.

[0117] The bill cost prediction device 100 for a logistics vehicle may also include one or more power supplies 14, one or more wired or wireless network interfaces 15, one or more input and output interfaces 16, and / or one or more operating systems 131, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Fig. 9 The device structure shown does not constitute a limitation on the bill cost prediction device 100 for logistics vehicles, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0118] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of a method for predicting bill costs for logistics vehicles.

[0119] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0120] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0121] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting bill costs of logistics vehicles, characterized in that: The bill cost analysis method comprises: Obtaining the driving data and cost data of the logistics vehicle, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the logistics vehicle's freight bill; The driving data and the cost data are pre-processed by using big data processing technology and uploaded to the blockchain; Based on the trained data enhancement model, the processed driving data and cost data are learned to obtain the best action strategy; Use the trained bill cost prediction model to predict the logistics vehicle bill costs for the best action strategy.

2. The method for predicting bill costs of logistics vehicles according to claim 1, characterized in that: Obtaining the driving data and cost data of logistics vehicles includes the following steps: Logistics vehicles are pre-equipped with GPS locators, fuel consumption sensors, and load sensors. The GPS locators are used to obtain the average speed, distance, and route of the vehicle. The fuel consumption sensors are used to monitor fuel consumption in real time. The load sensors are used to obtain the vehicle load. Retrieve weather data, traffic data, and real-time oil price data through Internet APIs; According to the freight bill of the logistics vehicle, the shuttle fee, freight bill quality control fine, fuel cost, toll and parking fee of the logistics vehicle are obtained, and the overall cost of the freight bill is calculated.

3. The method for predicting bill costs of logistics vehicles according to claim 1 is characterized in that: The driving data and the cost data are pre-processed by using big data processing technology and uploaded to the blockchain, including the following steps: The driving data and the cost data are cleaned, converted and loaded through ETL tools to finally form unified pre-processed data; Converting the preprocessed data into a format suitable for storage on a blockchain; Write smart contracts to define data storage and operation rules; Deploy the smart contract to the blockchain platform and call the method in the smart contract to upload the pre-processed data to the blockchain.

4. The method for predicting bill costs of logistics vehicles according to claim 1, characterized in that: Based on the trained data enhancement model, the processed logistics vehicle data is learned to obtain the best action strategy, including the following steps: Collect historical driving data and cost data of logistics vehicles, wherein the historical driving data includes vehicle type, average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data includes the overall cost of the logistics vehicle's freight bill; Standardize the historical driving data and cost-related data to ensure that data of different dimensions are comparable; Build a data enhancement learning model based on the enhancement learning algorithm; The data enhancement learning model is trained according to the historical driving data and cost-related data of the logistics vehicle, so that the data enhancement learning model learns to take corresponding actions in different states to obtain the maximum reward, thereby obtaining a trained data enhancement model; The driving data of the current logistics vehicle is input into the trained data enhancement model for learning, and the optimal action strategy for the vehicle in different states is output.

5. The method for predicting bill costs of logistics vehicles according to claim 4 is characterized in that: Building a data enhancement learning model based on the enhancement learning algorithm includes the following steps: A data reinforcement learning model is constructed based on the Q-Learning reinforcement learning algorithm, where the update formula of Q-Learning is: Q(s,a)←Q(s,a)+α[r+γa′maxQ(s′,a′)-Q(s,a)], where s and s′ represent the current state and the next state, respectively, a and a′ represent the current action and the possible actions in the next state, respectively, r is the immediate reward obtained after executing action a, α is the learning rate, and γ is the discount factor; The status, actions and rewards of the logistics vehicles in the data enhancement model are defined, wherein the status of the logistics vehicles is defined to include: the vehicle's starting position, the vehicle's terminal position, the vehicle type, the vehicle's load, and the vehicle's fuel consumption; the actions of the logistics vehicles are defined to include: the driving route and the average driving speed; and the reward obtained by the vehicle after taking a certain action is defined as: the reduced overall freight bill cost.

6. The method for predicting bill costs of logistics vehicles according to claim 1, characterized in that: The logistics vehicle bill expense trend is predicted using the trained neural network model for the best action strategy, including the following steps: Collect historical cost data of logistics vehicles under different action strategies; Construct an initial bill cost prediction model based on convolutional neural network; Using historical expense data as labels and different action strategies of logistics vehicles as feature data, the initial bill expense prediction model is trained, predicted values ​​are calculated through forward propagation, and weights are adjusted through back propagation to minimize the error between the predicted value and the actual value, thereby obtaining a trained bill expense prediction model; The optimal action strategy of the logistics vehicle is input into the trained bill cost prediction model, and the predicted bill cost of the logistics vehicle is output.

7. The method for predicting bill costs of logistics vehicles according to claim 1, characterized in that: Also includes the steps: Predict the total bill costs of logistics vehicles within the next month under the optimal action strategy based on the trained bill cost prediction model; If the total predicted logistics vehicle bill costs within the next month exceed the total budget, the logistics vehicle itinerary will be readjusted.

8. A bill cost prediction device for a logistics vehicle, characterized in that: include: An acquisition module is used to acquire the driving data and cost data of the logistics vehicle, wherein the driving data includes average driving speed, vehicle load, fuel consumption, driving distance, driving route, weather data, traffic condition data, and real-time oil price data, and the cost data is the overall cost of the freight bill of the logistics vehicle; A preprocessing module, used to preprocess the driving data and fee data using big data processing technology and upload them to the blockchain; A learning module is used to learn the processed driving data and cost data based on the trained data enhancement model to learn the best action strategy; The prediction module is used to predict the bill costs of logistics vehicles with the best action strategy by using the trained bill cost prediction model.

9. A bill cost prediction device for a logistics vehicle, characterized in that: comprising a memory and at least one processor, wherein the memory has computer-readable instructions stored therein; The at least one processor calls the computer-readable instructions in the memory to execute the various steps of the bill cost prediction method for a logistics vehicle as described in any one of claims 1-7.

10. A computer-readable storage medium having computer-readable instructions stored thereon, characterized in that: When the computer-readable instructions are executed by a processor, the various steps of the bill cost prediction method for a logistics vehicle as described in any one of claims 1-7 are implemented.

Images