Intelligent logistics scheduling optimization method and system based on AI technology

Abstract

The invention discloses an intelligent logistics scheduling optimization method and system based on an AI technology, and relates to the technical field of data traceability, and the method comprises the following steps: S1, obtaining real-time traffic data and distribution task information; s2, constructing a traffic prediction model based on the real-time traffic data, and predicting an emergent traffic problem occurring on a distribution path; and S3, adjusting a distribution route and a scheduling plan in combination with the distribution task information and a result of the traffic prediction model so as to avoid a predicted sudden traffic problem. According to the method, the traffic prediction model is constructed by acquiring the real-time traffic data, the sudden traffic problem occurring on the distribution path is predicted, and the distribution path and the scheduling plan are adjusted in combination with the distribution task information and the result of the traffic prediction model, so that the predicted sudden traffic problem is avoided, and the influence caused by the sudden traffic problem is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics scheduling, and specifically to a method and system for intelligent logistics scheduling optimization based on AI technology. Background Art

[0002] Intelligent logistics scheduling is the use of advanced technologies (such as artificial intelligence, big data, the Internet of Things, etc.) to optimize scheduling management in the logistics and transportation process, thereby improving efficiency, reducing costs and enhancing customer satisfaction.

[0003] However, in practical applications, intelligent logistics scheduling still has many shortcomings, including model complexity, data quality issues and poor dynamic adaptability. The shortcomings of poor dynamic adaptability, especially when facing sudden traffic problems, will lead to large errors in logistics delivery and time control, which will damage the logistics reputation. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the problems existing in the above-mentioned existing intelligent logistics scheduling optimization methods, the present invention provides the following technical solutions: an intelligent logistics scheduling optimization method based on AI technology, comprising: S1, obtaining real-time traffic data and distribution task information;

[0006] S2. Building a traffic prediction model based on the real-time traffic data and predicting sudden traffic problems on the delivery route;

[0007] S3. Adjust the delivery route and scheduling plan based on the delivery task information and the results of the traffic prediction model to avoid expected sudden traffic problems.

[0008] Preferably, the step S2 specifically includes:

[0009] Sa1. Based on neural network technology, learn historical traffic flow and weather patterns and train a preliminary traffic forecast model, including:

[0010] Construct training and validation sets based on the historical traffic flow and time series relationship;

[0011] Calculating the prediction error of the preliminary traffic prediction model on the training set according to a forward propagation algorithm of a neural network;

[0012] If the accuracy E of the model on the validation set is lower than the set accuracy P, the network structure is adjusted and retrained until E is not lower than P, at which point the network becomes the preliminary traffic prediction model;

[0013] During the training process, the stopping criterion is set as follows: if the model performance no longer improves significantly after the number of training iterations Q exceeds the predetermined threshold R, the model is considered to have converged and the training process ends;

[0014] Sa2, using the current real-time traffic flow data to update the relevant parameters in the preliminary traffic prediction model,

[0015] Specifically include:

[0016] After receiving new real-time traffic data Tn, the dynamic update mechanism of model parameters is started;

[0017] Calculating and updating the weight α based on the current real-time traffic flow speed V, reducing α when the accuracy of the model prediction decreases, and increasing α when it increases;

[0018] The new data is multiplied by the existing model parameters with the coefficient 1α, and then the relevant prediction parameters are updated by adding α and the current new real-time traffic parameters;

[0019] Determine whether the traffic prediction result after dynamic parameter update still meets the stability standard of threshold Y. If it meets the standard, execute the next process. If it does not meet the standard, continue to adjust the coefficient to find the best matching parameter value. Here, the stability standard is that the fluctuation of prediction accuracy Δ does not exceed 5;

[0020] Sa3, setting a probability threshold T for sudden traffic situations, and predicting traffic bottlenecks and congestion levels C based on the real-time traffic flow data and statistical information of traffic events, specifically including:

[0021] Obtain the number of vehicles N passing through the road section in unit time and the actual number of recorded traffic bottlenecks M;

[0022] The probability T of traffic incidents is calculated using the Bayesian formula, which is mathematically expressed as: T = P(M|N) = M / N*C, where C is the correction coefficient;

[0023] Determine whether the obtained T exceeds the warning value X (for example, T>X). If T exceeds the warning value X, it is determined that the road section may be congested or blocked, triggering the route adjustment strategy. Otherwise, the route is not adjusted and the original transportation scheduling is maintained;

[0024] If the new traffic speed and traffic volume after adjustment still cannot alleviate the congestion problem, the dynamic update is performed again and the judgment is repeated until T drops below the warning value X or reaches the maximum tolerance deviation threshold B, then the dynamic adjustment is terminated;

[0025] When the congestion level C exceeds the threshold T, it is considered that an emergency traffic situation will occur;

[0026] Sa4. Comprehensively consider various traffic influencing factors such as construction, accidents and bad weather to further optimize the preliminary traffic prediction model, so as to obtain more accurate prediction results.

[0027] Preferably, the specific method steps of using the Bayesian formula to predict traffic conditions in step Sa3 specifically include:

[0028] For a specific road intersection, obtain the historical time window t where traffic events frequently occur;

[0029] Establishing a statistical relationship using the historical ratio of the historical traffic flow N within the time window t to the number of traffic incidents M that actually occurred in history;

[0030] The ratio P(M) and the real-time traffic flow N are used to update the probability T of traffic congestion at the intersection in the time window. If T is still higher than the predetermined threshold V after the update, action is taken to adjust the route.

[0031] The formulaic judgment is: if T after the updated calculation is still greater than or equal to V, it is considered that major traffic obstacles may occur and measures need to be taken to reduce the risk level to the threshold level U or lower under normal conditions, then the line update process is terminated.

[0032] Preferably, the sudden traffic problems occurring on the delivery path in step S2 include:

[0033] Collect road construction information and accident report data;

[0034] Analyze historical traffic flow patterns to establish a baseline model;

[0035] Update model parameters based on current traffic congestion;

[0036] Use the updated traffic prediction model to predict road traffic conditions within a specific time period.

[0037] Preferably, the step of analyzing historical traffic flow patterns to establish a baseline model specifically includes:

[0038] Identify multiple factors that affect road accessibility based on historical traffic congestion conditions, including seasonality, time period, and holiday characteristics;

[0039] A statistical analysis framework is set up to calculate traffic change trends under different factors. When the correlation index is higher than the set threshold, the trend is added to the baseline.

[0040] For the key time points with significant impact, the traffic forecast value is calculated using the formula forecast traffic = (basic traffic x time correlation coefficient + holiday effect traffic);

[0041] After completing the above operations, a complete traffic flow prediction baseline is obtained.

[0042] Preferably, the flow prediction using the above formula at the key time point with significant impact also needs to include the following conditions:

[0043] Verify whether the current time is a holiday. If so, use the traffic model with holiday effect correction to predict traffic. That is, the prediction formula is changed to predicted traffic = (basic traffic x time holiday coefficient + holiday corrected traffic). The holiday corrected traffic is calculated through the additionally collected historical holiday traffic data, while the basic traffic is obtained based on the analysis of ordinary time points.

[0044] If it is not a holiday, the normal traffic flow analysis mode without holiday correction is used;

[0045] If the current time is a non-peak period at night, consider using a simplified algorithm to reduce computational complexity and save processing resources, such as simplifying the formula to predicted flow = (basic flow x night discount factor value);

[0046] Make sure to account for all time-specific factors when completing the above calculations and use them to update the overall baseline model.

[0047] Preferably, when verifying that the current time is in the nighttime non-peak period and using a simplified algorithm, the following steps are further included:

[0048] Confirm whether the actual traffic status at non-peak hours decreases abnormally. If so, use a more refined time discount rate to calculate the forecast;

[0049] The final correction value of the flow forecast is calculated by the formula of actual adjusted predicted flow = (basic flow measured deviation ratio) to adapt to the actual data changes monitored in real time. Here, the measured deviation ratio is the proportional difference between the current measured flow and the average flow during the expected peak period;

[0050] Determine whether the actual adjusted flow rate is lower than 80% of the historical average. If not, maintain the current simplified flow analysis plan unchanged and continue to monitor subsequent traffic dynamic information;

[0051] The monitoring system will automatically update and optimize the parameters and condition thresholds of the above steps regularly based on actual conditions.

[0052] Preferably, if the monitoring system finds that the actual flow rate in the current period is significantly less than the expected flow rate, further processing mechanisms are required, including:

[0053] When the ratio X of the actual measured basic flow to the expected peak average is less than 8 (i.e. the actual adjusted flow is lower than 80% of the historical average), measures are taken to provide real-time traffic information reminders and appropriately reduce the expected delivery speed to reduce possible delays;

[0054] Otherwise, it is considered that the current actual traffic conditions are close to the expected value range, and the original prediction model parameter configuration is kept unchanged without adjusting the distribution plan;

[0055] If X≤0.6 (the actual measured ratio is lower than 60% of the historical average), implement special emergency plans, such as adding spare delivery personnel and alternative delivery routes to ensure service continuity;

[0056] Ultimately, the initialization parameters and adjustment factors of the model for the next cycle will be dynamically determined based on the adjustment situation in the current stage.

[0057] An intelligent logistics scheduling optimization system based on AI technology, including:

[0058] Multiple collection terminals;

[0059] one or more processors;

[0060] A storage device for storing one or more programs:

[0061] When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 8.

[0062] Preferably, the acquisition terminal includes but is not limited to video collectors on various roads.

[0063] The present invention provides a method and system for intelligent logistics scheduling optimization based on AI technology. It has the following beneficial effects:

[0064] The present invention acquires real-time traffic data to construct a traffic prediction model and predicts sudden traffic problems on the delivery path. It adjusts the delivery route and scheduling plan based on the delivery task information and the results of the traffic prediction model to avoid the expected sudden traffic problems and reduce the impact of sudden traffic problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0066] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0067] An intelligent logistics scheduling optimization system based on AI technology, comprising: multiple collection terminals; one or more processors; a storage device for storing one or more programs: when the one or more programs are executed by the one or more processors.

[0068] The acquisition terminals include but are not limited to video collectors on various roads.

[0069] Embodiment 1

[0070] The present invention discloses a method and system for intelligent logistics scheduling optimization based on AI technology, and a method for intelligent logistics scheduling optimization based on AI technology, comprising:

[0071] S1. Obtain real-time traffic data and delivery task information;

[0072] S2. Building a traffic prediction model based on the real-time traffic data and predicting sudden traffic problems on the delivery route. The specific steps include:

[0073] Sa1. Based on neural network technology, learn historical traffic flow and weather patterns and train a preliminary traffic forecast model, including:

[0074] Construct training and validation sets based on the historical traffic flow and time series relationship;

[0075] Calculating the prediction error of the preliminary traffic prediction model on the training set according to a forward propagation algorithm of a neural network;

[0076] If the accuracy E of the model on the validation set is lower than the set accuracy P, the network structure is adjusted and retrained until E is not lower than P, at which point the network becomes the preliminary traffic prediction model;

[0077] During the training process, the stopping criterion is set as follows: if the model performance no longer improves significantly after the number of training iterations Q exceeds the predetermined threshold R, the model is considered to have converged and the training process ends;

[0078] Sa2, using the current real-time traffic flow data to update the relevant parameters in the preliminary traffic prediction model,

[0079] Specifically include:

[0080] After receiving new real-time traffic data Tn, the dynamic update mechanism of model parameters is started;

[0081] Calculating and updating the weight α based on the current real-time traffic flow speed V, reducing α when the accuracy of the model prediction decreases, and increasing α when it increases;

[0082] The new data is multiplied by the existing model parameters with the coefficient 1α, and then the relevant prediction parameters are updated by adding α and the current new real-time traffic parameters;

[0083] Determine whether the traffic prediction result after dynamic parameter update still meets the stability standard of threshold Y. If it meets the standard, execute the next process. If it does not meet the standard, continue to adjust the coefficient to find the best matching parameter value. Here, the stability standard is that the fluctuation of prediction accuracy Δ does not exceed 5;

[0084] Sa3, setting a probability threshold T for sudden traffic situations, and predicting traffic bottlenecks and congestion levels C based on the real-time traffic flow data and statistical information of traffic events, specifically including:

[0085] Obtain the number of vehicles N passing through the road section in unit time and the actual number of recorded traffic bottlenecks M;

[0086] The probability T of traffic incidents is calculated using the Bayesian formula, which is mathematically expressed as: T = P(M|N) = M / N*C, where C is the correction coefficient;

[0087] Determine whether the obtained T exceeds the warning value X (for example, T>X). If T exceeds the warning value X, it is determined that the road section may be congested or blocked, triggering the route adjustment strategy. Otherwise, the route is not adjusted and the original transportation scheduling is maintained;

[0088] If the new traffic speed and traffic volume after adjustment still cannot alleviate the congestion problem, the dynamic update is performed again and the judgment is repeated until T drops below the warning value X or reaches the maximum tolerance deviation threshold B, then the dynamic adjustment is terminated;

[0089] When the congestion level C exceeds the threshold T, it is considered that an emergency traffic situation will occur;

[0090] Sa4. Comprehensively consider various traffic influencing factors such as construction, accidents and bad weather to further optimize the preliminary traffic prediction model, so as to obtain more accurate prediction results.

[0091] The specific method steps of using the Bayesian formula to predict traffic conditions in step Sa3 specifically include:

[0092] For a specific road intersection, obtain the historical time window t where traffic events frequently occur;

[0093] Establishing a statistical relationship using the historical ratio of the historical traffic flow N within the time window t to the number of traffic incidents M that actually occurred in history;

[0094] The ratio P(M) and the real-time traffic flow N are used to update the probability T of traffic congestion at the intersection in the time window. If T is still higher than the predetermined threshold V after the update, action is taken to adjust the route.

[0095] The formulaic judgment is: if T after the updated calculation is still greater than or equal to V, it is considered that major traffic obstacles may occur and measures need to be taken to reduce the risk level to the threshold level U or lower under normal conditions, then the line update process is terminated.

[0096] S3. Adjust the delivery route and scheduling plan based on the delivery task information and the results of the traffic prediction model to avoid expected sudden traffic problems.

[0097] The sudden traffic problems occurring on the delivery path in step S2 include:

[0098] Collect road construction information and accident report data;

[0099] Analyze historical traffic flow patterns to establish a baseline model;

[0100] Update model parameters based on current traffic congestion;

[0101] Use the updated traffic prediction model to predict road traffic conditions within a specific time period.

[0102] The steps of analyzing historical traffic flow patterns and establishing a baseline model specifically include:

[0103] Identify multiple factors that affect road accessibility based on historical traffic congestion conditions, including seasonality, time period, and holiday characteristics;

[0104] A statistical analysis framework is set up to calculate traffic change trends under different factors. When the correlation index is higher than the set threshold, the trend is added to the baseline.

[0105] For the key time points with significant impact, the traffic forecast value is calculated using the formula forecast traffic = (basic traffic x time correlation coefficient + holiday effect traffic);

[0106] After completing the above operations, a complete traffic flow prediction baseline is obtained.

[0107] When using the above formula to predict traffic flow at a key time point with significant impact, the following conditions must also be included:

[0108] Verify whether the current time is a holiday. If so, use the traffic model with holiday effect correction to predict traffic. That is, the prediction formula is changed to predicted traffic = (basic traffic x time holiday coefficient + holiday corrected traffic). The holiday corrected traffic is calculated through the additionally collected historical holiday traffic data, while the basic traffic is obtained based on the analysis of ordinary time points.

[0109] If it is not a holiday, the normal traffic flow analysis mode without holiday correction is used.

[0110] Embodiment 2

[0111] The present invention discloses a method and system for intelligent logistics scheduling optimization based on AI technology, and a method for intelligent logistics scheduling optimization based on AI technology, comprising:

[0112] S1. Obtain real-time traffic data and delivery task information;

[0113] S2. Building a traffic prediction model based on the real-time traffic data and predicting sudden traffic problems on the delivery route. The specific steps include:

[0114] Sa1. Based on neural network technology, learn historical traffic flow and weather patterns and train a preliminary traffic forecast model, including:

[0115] Construct training and validation sets based on the historical traffic flow and time series relationship;

[0116] Calculating the prediction error of the preliminary traffic prediction model on the training set according to a forward propagation algorithm of a neural network;

[0117] If the accuracy E of the model on the validation set is lower than the set accuracy P, the network structure is adjusted and retrained until E is not lower than P, at which point the network becomes the preliminary traffic prediction model;

[0118] During the training process, the stopping criterion is set as follows: if the model performance no longer improves significantly after the number of training iterations Q exceeds the predetermined threshold R, the model is considered to have converged and the training process ends;

[0119] Sa2, using the current real-time traffic flow data to update the relevant parameters in the preliminary traffic prediction model,

[0120] Specifically include:

[0121] After receiving new real-time traffic data Tn, the dynamic update mechanism of model parameters is started;

[0122] Calculating and updating the weight α based on the current real-time traffic flow speed V, reducing α when the accuracy of the model prediction decreases, and increasing α when it increases;

[0123] The new data is multiplied by the existing model parameters with the coefficient 1α, and then the relevant prediction parameters are updated by adding α and the current new real-time traffic parameters;

[0124] Determine whether the traffic prediction result after dynamic parameter update still meets the stability standard of threshold Y. If it meets the standard, execute the next process. If it does not meet the standard, continue to adjust the coefficient to find the best matching parameter value. Here, the stability standard is that the fluctuation of prediction accuracy Δ does not exceed 5;

[0125] Sa3, setting a probability threshold T for sudden traffic situations, and predicting traffic bottlenecks and congestion levels C based on the real-time traffic flow data and statistical information of traffic events, specifically including:

[0126] Obtain the number of vehicles N passing through the road section in unit time and the actual number of recorded traffic bottlenecks M;

[0127] The probability T of traffic incidents is calculated using the Bayesian formula, which is mathematically expressed as: T = P(M|N) = M / N*C, where C is the correction coefficient;

[0128] Determine whether the obtained T exceeds the warning value X (for example, T>X). If T exceeds the warning value X, it is determined that the road section may be congested or blocked, triggering the route adjustment strategy. Otherwise, the route is not adjusted and the original transportation scheduling is maintained;

[0129] If the new traffic speed and traffic volume after adjustment still cannot alleviate the congestion problem, the dynamic update is performed again and the judgment is repeated until T drops below the warning value X or reaches the maximum tolerance deviation threshold B, then the dynamic adjustment is terminated;

[0130] When the congestion level C exceeds the threshold T, it is considered that an emergency traffic situation will occur;

[0131] Sa4. Comprehensively consider various traffic influencing factors such as construction, accidents and bad weather to further optimize the preliminary traffic prediction model, so as to obtain more accurate prediction results.

[0132] The specific method steps of using the Bayesian formula to predict traffic conditions in step Sa3 specifically include:

[0133] For a specific road intersection, obtain the historical time window t where traffic events frequently occur;

[0134] Establishing a statistical relationship using the historical ratio of the historical traffic flow N within the time window t to the number of traffic incidents M that actually occurred in history;

[0135] The ratio P(M) and the real-time traffic flow N are used to update the probability T of traffic congestion at the intersection in the time window. If T is still higher than the predetermined threshold V after the update, action is taken to adjust the route.

[0136] The formulaic judgment is: if T after the updated calculation is still greater than or equal to V, it is considered that major traffic obstacles may occur and measures need to be taken to reduce the risk level to the threshold level U or lower under normal conditions, then the line update process is terminated.

[0137] S3. Adjust the delivery route and scheduling plan based on the delivery task information and the results of the traffic prediction model to avoid expected sudden traffic problems.

[0138] The sudden traffic problems occurring on the delivery path in step S2 include:

[0139] Collect road construction information and accident report data;

[0140] Analyze historical traffic flow patterns to establish a baseline model;

[0141] Update model parameters based on current traffic congestion;

[0142] Use the updated traffic prediction model to predict road traffic conditions within a specific time period.

[0143] The steps of analyzing historical traffic flow patterns and establishing a baseline model specifically include:

[0144] Identify multiple factors that affect road accessibility based on historical traffic congestion conditions, including seasonality, time period, and holiday characteristics;

[0145] A statistical analysis framework is set up to calculate traffic change trends under different factors. When the correlation index is higher than the set threshold, the trend is added to the baseline.

[0146] For the key time points with significant impact, the traffic forecast value is calculated using the formula forecast traffic = (basic traffic x time correlation coefficient + holiday effect traffic);

[0147] After completing the above operations, a complete traffic flow prediction baseline is obtained.

[0148] When using the above formula to predict traffic flow at a key time point with significant impact, the following conditions must also be included:

[0149] If the current time is a non-peak period at night, consider using a simplified algorithm to reduce computational complexity and save processing resources, such as simplifying the formula to predicted flow = (basic flow x night discount factor value);

[0150] Make sure to account for all time-specific factors when completing the above calculations and use them to update the overall baseline model.

[0151] When verifying that the current time is in the nighttime non-peak period and using the simplified algorithm, the following steps are also required:

[0152] Confirm whether the actual traffic status at non-peak hours decreases abnormally. If so, use a more refined time discount rate to calculate the forecast;

[0153] The final correction value of the flow forecast is calculated by the formula of actual adjusted predicted flow = (basic flow measured deviation ratio) to adapt to the actual data changes monitored in real time. Here, the measured deviation ratio is the proportional difference between the current measured flow and the average flow during the expected peak period;

[0154] Determine whether the actual adjusted flow rate is lower than 80% of the historical average. If not, maintain the current simplified flow analysis plan unchanged and continue to monitor subsequent traffic dynamic information;

[0155] The monitoring system will automatically update and optimize the parameters and condition thresholds of the above steps regularly based on actual conditions.

[0156] If the monitoring system finds that the actual flow rate in the current period is significantly less than the expected flow rate, further processing mechanisms are required, including:

[0157] When the ratio X of the actual measured basic flow to the expected peak average is less than 8 (i.e. the actual adjusted flow is lower than 80% of the historical average), measures are taken to provide real-time traffic information reminders and appropriately reduce the expected delivery speed to reduce possible delays;

[0158] Otherwise, it is considered that the current actual traffic conditions are close to the expected value range, and the original prediction model parameter configuration is kept unchanged without adjusting the distribution plan;

[0159] If X≤0.6 (the actual measured ratio is lower than 60% of the historical average), implement special emergency plans, such as adding spare delivery personnel and alternative delivery routes to ensure service continuity;

[0160] Ultimately, the initialization parameters and adjustment factors of the model for the next cycle will be dynamically determined based on the adjustment situation in the current stage.

[0161] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for intelligent logistics scheduling optimization based on AI technology, characterized in that: The following steps are involved: S1. Obtain real-time traffic data and delivery task information; S2. Building a traffic prediction model based on the real-time traffic data and predicting sudden traffic problems on the delivery route; S3. Adjust the delivery route and scheduling plan based on the delivery task information and the results of the traffic prediction model to avoid expected sudden traffic problems.

2. The method for intelligent logistics scheduling optimization based on AI technology according to claim 1 is characterized in that: The step S2 specifically includes: Sa1. Based on neural network technology, we learn historical traffic flow and weather patterns and train a preliminary traffic forecast model, including: Construct training and validation sets based on the historical traffic flow and time series relationship; Calculating the prediction error of the preliminary traffic prediction model on the training set according to a forward propagation algorithm of a neural network; If the accuracy E of the model on the validation set is lower than the set accuracy P, the network structure is adjusted and retrained until E is not lower than P, at which point the network becomes the preliminary traffic prediction model; During the training process, the stopping criterion is set as follows: if the model performance no longer improves significantly after the number of training iterations Q exceeds the predetermined threshold R, the model is considered to have converged and the training process ends; Sa2. Use the current real-time traffic flow data to update the relevant parameters in the preliminary traffic prediction model, including: After receiving new real-time traffic data Tn, the dynamic update mechanism of model parameters is started; Calculating and updating the weight α based on the current real-time traffic flow speed V, reducing α when the accuracy of the model prediction decreases, and increasing α when it increases; The new data is multiplied by the existing model parameters with the coefficient 1α, and then the relevant prediction parameters are updated by adding α and the current new real-time traffic parameters; Determine whether the traffic prediction result after dynamic parameter update still meets the stability standard of threshold Y. If it meets the standard, execute the next process. If it does not meet the standard, continue to adjust the coefficient to find the best matching parameter value. Here, the stability standard is that the fluctuation of prediction accuracy Δ does not exceed 5; Sa3, setting a probability threshold T for sudden traffic situations, and predicting traffic bottlenecks and congestion levels C based on the real-time traffic flow data and statistical information of traffic events, specifically including: Obtain the number of vehicles N passing through the road section in unit time and the actual number of recorded traffic bottlenecks M; The probability T of traffic incidents is calculated using the Bayesian formula, which is mathematically expressed as: T = P(M|N) = M / N*C, where C is the correction coefficient; Determine whether the obtained T exceeds the warning value X (for example, T>X). If T exceeds the warning value X, it is determined that the road section may be congested or blocked, triggering the route adjustment strategy. Otherwise, the route is not adjusted and the original transportation scheduling is maintained; If the new traffic speed and traffic volume after adjustment still cannot alleviate the congestion problem, the dynamic update is performed again and the judgment is repeated until T drops below the warning value X or reaches the maximum tolerance deviation threshold B, then the dynamic adjustment is terminated; When the congestion level C exceeds the threshold T, it is considered that an emergency traffic situation will occur; Sa4. Comprehensively consider various traffic influencing factors such as construction, accidents and bad weather to further optimize the preliminary traffic prediction model, so as to obtain more accurate prediction results.

3. According to claim 2, an intelligent logistics scheduling optimization method based on AI technology is characterized in that: The specific method steps of using the Bayesian formula to predict traffic conditions in step Sa3 specifically include: For a specific road intersection, obtain the historical time window t where traffic events frequently occur; Establishing a statistical relationship using the historical ratio of the historical traffic flow N within the time window t to the number of traffic incidents M that actually occurred in history; The ratio P(M) and the real-time traffic flow N are used to update the probability T of traffic congestion at the intersection in the time window. If T is still higher than the predetermined threshold V after the update, action is taken to adjust the route. The formulaic judgment is: if T after the updated calculation is still greater than or equal to V, it is considered that major traffic obstacles may occur and measures need to be taken to reduce the risk level to the threshold level U or lower under normal conditions, then the line update process is terminated.

4. According to the AI-based intelligent logistics scheduling optimization method of claim 1, it is characterized in that: The sudden traffic problems occurring on the delivery path in step S2 include: Collect road construction information and accident report data; Analyze historical traffic flow patterns to establish a baseline model; Update model parameters based on current traffic congestion; Use the updated traffic prediction model to predict road traffic conditions within a specific time period.

5. According to claim 4, the intelligent logistics scheduling optimization method based on AI technology is characterized in that: The steps of analyzing historical traffic flow patterns and establishing a baseline model specifically include: Identify multiple factors that affect road accessibility based on historical traffic congestion conditions, including seasonality, time period, and holiday characteristics; A statistical analysis framework is set up to calculate traffic change trends under different factors. When the correlation index is higher than the set threshold, the trend is added to the baseline. For the key time points with significant impact, the traffic forecast value is calculated using the formula forecast traffic = (basic traffic x time correlation coefficient + holiday effect traffic); After completing the above operations, a complete traffic flow prediction baseline is obtained.

6. The intelligent logistics scheduling optimization method based on AI technology according to claim 5 is characterized in that: When using the above formula to predict traffic flow at a key time point with significant impact, the following conditions must also be included: Verify whether the current time is a holiday. If so, use the traffic model with holiday effect correction to predict traffic. That is, the prediction formula is changed to predicted traffic = (basic traffic x time holiday coefficient + holiday corrected traffic). The holiday corrected traffic is calculated through the additionally collected historical holiday traffic data, while the basic traffic is obtained based on the analysis of ordinary time points. If it is not a holiday, the normal traffic flow analysis mode without holiday correction is used; If the current time is a non-peak period at night, consider using a simplified algorithm to reduce computational complexity and save processing resources, such as simplifying the formula to predicted flow = (basic flow x night discount factor value); Make sure to account for all time-specific factors when completing the above calculations and use them to update the overall baseline model.

7. The intelligent logistics scheduling optimization method based on AI technology according to claim 6 is characterized in that: When verifying that the current time is in the nighttime non-peak period and using the simplified algorithm, the following steps are also required: Confirm whether the actual traffic status at non-peak hours decreases abnormally. If so, use a more refined time discount rate to calculate the forecast; The final correction value of the flow forecast is calculated by the formula of actual adjusted predicted flow = (basic flow measured deviation ratio) to adapt to the actual data changes monitored in real time. Here, the measured deviation ratio is the proportional difference between the current measured flow and the average flow during the expected peak period; Determine whether the actual adjusted flow rate is lower than 80% of the historical average. If not, maintain the current simplified flow analysis plan unchanged and continue to monitor subsequent traffic dynamic information; The monitoring system will automatically update and optimize the parameters and condition thresholds of the above steps regularly based on actual conditions.

8. The intelligent logistics scheduling optimization method based on AI technology according to claim 2 is characterized in that: If the monitoring system finds that the actual flow rate in the current period is significantly less than the expected flow rate, further processing mechanisms are required, including: When the ratio X of the actual measured basic flow to the expected peak average is less than 8 (i.e. the actual adjusted flow is lower than 80% of the historical average), measures are taken to provide real-time traffic information reminders and appropriately reduce the expected delivery speed to reduce possible delays; Otherwise, it is considered that the current actual traffic conditions are close to the expected value range, and the original prediction model parameter configuration is kept unchanged without adjusting the distribution plan; If X≤0.6 (the actual measured ratio is lower than 60% of the historical average), implement special emergency plans, such as adding spare delivery personnel and alternative delivery routes to ensure service continuity; Ultimately, the initialization parameters and adjustment factors of the model for the next cycle will be dynamically determined based on the adjustment situation in the current stage.

9. An intelligent logistics scheduling optimization system based on AI technology, characterized in that: include: Multiple collection terminals; one or more processors; A storage device for storing one or more programs: When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 8.

10. The intelligent logistics scheduling optimization system based on AI technology according to claim 9 is characterized in that: The acquisition terminals include but are not limited to video collectors on various roads.

Images