Traffic flow prediction method, system and model based on block data fusion ARIMA model

Abstract

The invention provides a traffic flow prediction method, system and model based on a block data fusion ARIMA model. The method comprises the following steps: carrying out data reconstruction and feature extraction on preprocessed data; combining a feature extraction result with the original data to form a plurality of data blocks; training the block data fusion ARIMA model on the basis of the data blocks; and carrying out residual analysis on the trained block data fusion ARIMA model, applying a residual to a random forest algorithm for modeling, and combining a residual prediction value obtained by the random forest algorithm with a prediction value of the ARIMA model to predict the traffic flow. According to the method, random forest algorithm modeling is carried out on the residual error of the ARIMA model, the defects of the ARIMA model are compensated by utilizing the strong capability of the random forest, and a complex mode and a nonlinear relationship in data can be better processed, so that the accuracy and reliability of prediction are improved.

Description

Technical Field

[0001] The present application belongs to the field of traffic information technology, and in particular relates to a traffic flow prediction method, system and model based on block data fusion ARIMA model. Background Art

[0002] As a core component of intelligent transportation systems, traffic flow forecasting is of great significance in alleviating urban traffic congestion, improving road efficiency, and reducing traffic accidents. With the acceleration of urbanization, traffic flow data presents more complex spatiotemporal characteristics, and traditional forecasting methods can no longer meet the growing demand for forecasting accuracy.

[0003] Existing technologies mainly use time series analysis methods, especially the ARIMA model, which is widely used in traffic flow prediction due to its effectiveness in processing linear time series data. However, the ARIMA model performs poorly when processing data with nonlinear characteristics and has difficulty capturing the spatiotemporal dependence of traffic flow. Summary of the invention

[0004] In view of the limitations of existing methods, a traffic flow prediction method, system and model based on block data fusion ARIMA model are proposed. This method divides the data such as traffic flow, road conditions, weather and special events into multiple data blocks, uses data reconstruction technology to extract the intrinsic characteristics of each data block, and then effectively integrates the characteristic information of each data block through a fusion mechanism to form a new block data, which is input into the ARIMA model for prediction. This method can not only improve the prediction accuracy, but also enhance the model's ability to capture nonlinearity and spatiotemporal dependencies.

[0005] To achieve the above objectives, the present application provides a traffic flow prediction method based on block data fusion ARIMA model, which specifically includes: S1: Preprocess the collected raw data, and perform data reconstruction and feature extraction on the preprocessed data; S2: combining the feature extraction result with the original data to form multiple data blocks; S3: constructing a block data fusion ARIMA model, and training the block data fusion ARIMA model based on the data block; S4: Perform residual analysis on the trained block data fusion ARIMA model, apply the residual to the random forest algorithm for modeling, capture nonlinear and complex interaction effects, and combine the residual prediction values ​​obtained by the random forest algorithm with the prediction values ​​of the ARIMA model to predict traffic flow.

[0006] The step S1 performs data reconstruction and feature extraction on the preprocessed data, specifically including: Reconstruct the time series data into a high-order tensor using multi-path delay embedding transform technology; A high-order tensor is decomposed into the product of multiple low-order tensors, wherein the low-order tensors represent the time trend, periodic changes, and abnormal fluctuations of the data.

[0007] The step S3 constructs a block data fusion ARIMA model, specifically including: Integrate the multiple data blocks into the ARIMA model, determine the order q of the MA part through the ACF graph; determine the order p of the AR part through the PACF graph; The order d is determined by using second-order backward difference; Use the AIC criterion to select the optimal ARIMA model parameters; Estimate the coefficients of the AR and MA parts by the MLE method, perform statistical hypothesis tests on these coefficients, and screen for significant coefficients; The model is adjusted according to the significance and refitted until a block data fusion ARIMA model is obtained.

[0008] The step S3 further includes training the block data fusion ARIMA model based on the data block: Use statsmodels to import the necessary libraries and use pandas to read the data block data; Train the ARIMA model according to the order q, order p and order d, and calculate the MAPE and R2 error indicators; Use TimeSeriesSplit to perform cross-validation of time series, evaluate the performance of the block data fusion ARIMA model in different time periods, and retrain the model after finding the optimal parameters.

[0009] The residual prediction value obtained by the random forest algorithm is combined with the prediction value of the ARIMA model to predict traffic flow using the following formula: ; in, is the combined prediction result, is a combined prediction model, is the prediction result of the ARIMA model, is the residual partial derivative of the random forest prediction, For time.

[0010] The preprocessing of the collected raw data specifically includes: Traffic flow, road conditions, weather and special events are collected as raw data, and the collected raw data are divided into multiple initial data blocks according to time and space. The divided initial data blocks are then subjected to data cleaning, missing value processing and outlier detection preprocessing.

[0011] To achieve the above purpose, the present application also provides a traffic flow prediction system based on block data fusion ARIMA model, the prediction system comprising: Data collection unit, which collects traffic flow, road conditions, weather and special events as raw data; The first processing unit preprocesses the collected raw data, and performs data reconstruction and feature extraction on the preprocessed data; The second processing unit constructs a block data fusion ARIMA model, trains and optimizes the block data fusion ARIMA model based on the data block, and obtains an optimal block data fusion ARIMA model; The computing unit performs traffic flow prediction based on the optimal block data and the ARIMA model.

[0012] Wherein, the second processing unit specifically includes: The model building module determines the hyperparameters (p, d, q) required for the ARIMA model by analyzing the characteristics of ACF and PACF, and then selects the optimal model parameters through the AIC value to build the block data fusion ARIMA model; The model optimization module estimates the coefficients of the AR and MA parts through the MLE method, performs statistical hypothesis tests on these coefficients, and screens significant coefficients; adjusts the model according to the significance, and refits until the optimized block data fusion ARIMA model is obtained.

[0013] To achieve the above objectives, the present application also provides a block data fusion ARIMA model, which is a block data fusion ARIMA model in a traffic flow prediction method based on the block data fusion ARIMA model as described above.

[0014] To achieve the above objectives, the present application also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by a control processor, the traffic flow prediction method based on the block data fusion ARIMA model as described above is implemented.

[0015] The present invention collects traffic flow, road conditions, weather, special events and other data through multiple channels, performs data preprocessing; reconstructs the preprocessed data into a high-order tensor, decomposes and extracts key features, and combines them to form new block data; integrates the new block data into an ARIMA model, performs model training and verification, and models the residuals through a random forest algorithm, and applies the optimized ARIMA combined model to traffic flow prediction to improve the accuracy and reliability of the prediction.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention can improve the prediction ability of traffic flow time series by combining the autoregression, difference and sliding average characteristics of time series by applying data blocks to ARIMA model. The residual of ARIMA model is modeled by random forest algorithm, and the powerful ability of random forest is used to compensate for the shortcomings of ARIMA model, which can better handle complex patterns and nonlinear relationships in data, thereby improving the accuracy and reliability of prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a system architecture diagram of a traffic flow prediction method based on block data fusion ARIMA model in one embodiment of the present application.

[0018] Figure 2 This is a schematic diagram of a traffic flow prediction method based on block data fusion ARIMA model in one embodiment of the present application.

[0019] Figure 3 This is a distribution diagram of MAPE and R2 error indicators of ARIMA single prediction results in one embodiment of the present application.

[0020] Figure 4 This is a distribution diagram of MAPE and R2 error indicators of the combined model prediction results in one embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme will be clearly and completely described below in conjunction with the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application. Embodiment 1

[0022] As attached Figure 1-2 As shown, in order to solve the above technical problems, the present application provides a traffic flow prediction method based on block data fusion ARIMA model, which specifically includes: S1: Preprocess the collected raw data, and perform data reconstruction and feature extraction on the preprocessed data; S2: combining the feature extraction result with the original data to form multiple data blocks; S3: constructing a block data fusion ARIMA model, and training the block data fusion ARIMA model based on the data block; S4: Perform residual analysis on the trained block data fusion ARIMA model, apply the residual to the random forest algorithm for modeling, capture nonlinear and complex interaction effects, and combine the residual prediction values ​​obtained by the random forest algorithm with the prediction values ​​of the ARIMA model to predict traffic flow.

[0023] Preferably, the collected raw data can be collected through multiple channels such as traffic flow, road conditions, weather and special events. In this embodiment, the collected data is further divided according to time and space to form multiple data blocks, and the divided data blocks are pre-processed such as data cleaning, missing value processing, and outlier detection.

[0024] Traffic flow and road condition data are derived from historical traffic monitoring records provided by the Urban Traffic Management Bureau. Meteorological data for the corresponding time period are also collected, including rainfall, temperature, wind speed, etc., as well as special events such as holidays and large-scale events that may cause abnormal fluctuations in traffic flow. All data are strictly quality-controlled to ensure their accuracy and completeness.

[0025] The original data is divided according to the time dimension and the space dimension to form multiple data blocks with logical and geographical significance, but it is not limited to this.

[0026] After the data blocks are divided, data cleaning is performed, including removing duplicate records, correcting data entry errors, unifying data formats, etc., to ensure data consistency and accuracy.

[0027] Identify and process missing values ​​in the data based on data cleaning, fill them with the mean, median or mode, or use more complex interpolation methods such as linear interpolation or polynomial interpolation. Outliers may be caused by data entry errors or measurement errors. Perform outlier detection on the data to improve the reliability of the data, identify and process outliers. After processing missing values ​​and outliers, standardize or normalize the data to eliminate the impact of different dimensions and magnitudes, and ensure the balance of various features during model training.

[0028] The preprocessed time series data is reconstructed into a high-order tensor, the reconstructed tensor is decomposed to extract key features, and the decomposed features are combined with the original time series data to form new block data.

[0029] In this embodiment, the preprocessed data is reconstructed into a high-order tensor through the multi-path delay embedding transform technology (MDT), which is implemented by the sliding window method, taking the continuous time point data as a slice of the tensor, mining the deep-level features of the time series data, and integrating the different dimensions of the time series data (such as time, space, category, etc.) into a unified data framework through reconstruction. The specific implementation process is as follows: (1) Define embedding parameters Choose the delay and embedding dimensions: Determine the length of each embedding vector (usually the size of the time window) and the delay step size. The embedding dimension is determined by the length of the time window and the sliding step size.

[0030] (2) Use the sliding window method to generate multiple time slices. For example, if the time series length is T, the time window size is W, and the sliding step size is S, then (T−W) / S+1 time slices are generated.

[0031] Each slice: Each time slice can be represented as an embedding vector. Assuming the time length of each slice is W, the shape of each slice is (W, F), where F is the number of features.

[0032] (3) All generated time slices are stacked sequentially into a high-order tensor. Assume that N slices are generated, each with a shape of (W, F), and the final tensor shape is (N, W, F).

[0033] Tensor decomposition technology is used to identify and extract key features in the data, decomposing high-order tensors into the product of several low-order tensors. Each low-order tensor represents different aspects of the data, such as time trends, periodic changes, abnormal fluctuations, etc. For example: CP decomposition: decompose the tensor X into the sum of several rank tensors: ; Among them, a r , b r and c r are the factor matrices of time, features, and categories, respectively, and R is the rank of the decomposition.

[0034] In another embodiment, Tucker decomposition may also be used: the tensor X is decomposed into a core tensor G and a factor matrix: ; Among them, A, B, and C are factor matrices, and the core tensor G represents the main features of the data.

[0035] In this embodiment, the key features obtained by decomposition are combined with the original time series data to form new block data. The new block data not only retains the integrity of the original data, but also increases the dimension and depth of the data, which is more suitable for multi-dimensional analysis and processing complex machine learning tasks.

[0036] The step S3 constructs a block data fusion ARIMA model, specifically including: The multiple data blocks are integrated into the ARIMA model, and the order q of the MA part is determined by the ACF diagram; the order p of the AR part is determined by the PACF diagram; wherein, for the AR part (p), the number of significant peaks in the PACF diagram corresponds to the value of p, and for the MA part (q), the number of significant peaks in the ACF diagram corresponds to the value of q.

[0037] The order d is determined by using the second-order backward difference; the second-order backward difference is used in this model Get better time series stationarity.

[0038] The AIC criterion is used to select the optimal ARIMA model parameters, where the smaller the AIC value, the better the model.

[0039] Estimate the coefficients of the AR and MA parts by the maximum likelihood estimation (MLE) method, find a set of parameters that maximizes the logarithm of the probability density function of the observed data, perform statistical hypothesis tests on these coefficients, and screen for significant coefficients; In this embodiment, the coefficients of the AR and MA parts in the model are estimated by the MLE method, which can be expressed as: ; where ϵt is the white noise error term, ϕ i is the coefficient of the AR part, θ j are the coefficients of the MA part.

[0040] Construct the likelihood function: ; Where ϵ^t is the residual calculated based on the AR and MA coefficients, and σ2 is the variance of the residual.

[0041] Maximize the log-likelihood function: In this embodiment, a numerical optimization method (such as gradient descent, Newton's method, etc.) can be used to find the AR and MA coefficients that maximize the log-likelihood function. Specifically, the ARIMA function in the statsmodels library in Python or the arima function in R can be used for MLE estimation.

[0042] Statistical hypothesis tests are performed on the AR and MA coefficients to assess whether each coefficient is significant: Among them, the null hypothesis H0: the coefficient is equal to zero, indicating that the coefficient does not contribute significantly to the model.

[0043] Alternative hypothesis H1: The coefficient is not equal to zero, indicating that the coefficient contributes significantly to the model.

[0044] Preferably, a t-test statistic is used to assess the significance of each coefficient: ; in is the MLE estimate of the coefficient, is the standard error of the coefficient.

[0045] The model is adjusted according to the significance and refitted until a block data fusion ARIMA model is obtained.

[0046] In this embodiment, the estimated parameters are statistically tested to determine whether they are significantly different from zero. Non-significant parameters may indicate that the model is overfitting and should be adjusted. If some parameters are found to be insignificant, these parameters can be set to 0, thereby simplifying the model, helping to avoid overfitting, and possibly improving the predictive performance of the model.

[0047] The step S3 further includes training the block data fusion ARIMA model based on the data block: Use statsmodels to import the necessary libraries and use pandas to read the data block data; According to the order q, order p and order d, the ARIMA model is trained and the function is defined. Train the ARIMA model and calculate the MAPE and R2 error indicators; Figure 3-4 As shown in the figure, the forecast results of the combined model are more stable and accurate than the MAPE and R2 error indicators of ARIMA's separate forecast.

[0048] in is the prediction result of the ARIMA model, is the ARIMA forecasting model, is the traffic flow data, For rainfall data, is the temperature data, The wind speed data is used; TimeSeriesSplit is used to perform cross-validation of the time series, evaluate the performance of the block data fusion ARIMA model in different time periods, and retrain the model after finding the optimal parameters.

[0049] On this basis, the model is retrained using the best parameters found, and validation and parameter optimization are continued. The parameter search range is adjusted according to actual conditions. ,in is the residual, is the residual extraction function, is the actual value, is the prediction result of the ARIMA model, time; use the random forest algorithm to train and predict the residual ,in For random forest prediction residual results, For the random forest prediction residual model, is the data frame conversion function, is time; the predicted value of the ARIMA model is combined with the residual predicted value obtained by the random forest algorithm Predict traffic flow, The model described in the present invention can be applied to traffic flow prediction. In combination with block data technology, the optimized ARIMA combination model can be applied to traffic flow prediction to achieve accurate prediction of traffic flow in different time periods and sections. The optimized ARIMA combination model can adapt to the dynamic changes of traffic flow. In combination with block data technology, the model can simultaneously consider multiple factors affecting traffic flow and achieve multi-dimensional traffic flow prediction; by block segmenting the traffic data of different sections, the model can provide customized prediction results for each section, helping traffic management departments to more accurately control and plan traffic; dynamically adjust the timing of traffic lights, release traffic information, etc., so as to effectively alleviate traffic pressure and improve road capacity. Embodiment 2

[0050] The present application also provides a traffic flow prediction system based on block data fusion ARIMA model, the prediction system comprising: Data collection unit, which collects traffic flow, road conditions, weather and special events as raw data; The first processing unit preprocesses the collected raw data, and performs data reconstruction and feature extraction on the preprocessed data; The second processing unit constructs a block data fusion ARIMA model, trains and optimizes the block data fusion ARIMA model based on the data block, and obtains an optimal block data fusion ARIMA model; The computing unit performs traffic flow prediction based on the optimal block data and the ARIMA model.

[0051] Wherein, the second processing unit specifically includes: The model building module determines the hyperparameters (p, d, q) required for the ARIMA model by analyzing the characteristics of ACF and PACF, and then selects the optimal model parameters through the AIC value to build the block data fusion ARIMA model; The model optimization module estimates the coefficients of the AR and MA parts through the MLE method, performs statistical hypothesis tests on these coefficients, and screens significant coefficients; adjusts the model according to the significance, and refits until the optimized block data fusion ARIMA model is obtained. Embodiment 3

[0052] The present application also provides a block data fusion ARIMA model, which is the block data fusion ARIMA model in the traffic flow prediction method based on the block data fusion ARIMA model as described above.

[0053] In this embodiment, the block data fusion ARIMA model fuses different data sources or data features through block data, thereby improving the prediction accuracy of the model. Embodiment 4

[0054] The present application also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by a control processor, the traffic flow prediction method based on the block data fusion ARIMA model as described above is implemented.

[0055] It should be noted that the storage medium in this embodiment includes computer programs, configuration files, user data, etc. In the anti-peeping display control device, the storage medium can store computer programs for controlling pixel units, lens groups, and switch control units. The storage medium must be computer readable, which means that the computer can access the data in the storage medium. For example, the storage medium can be a hard disk, a solid-state hard disk, a USB flash drive, an optical disk, etc.

[0056] The computer program on the storage medium can be executed by the processor. In the anti-peeping display control device, the processor can read the program on the storage medium and control the brightness and color of the pixel unit, adjust the position and shape of the lens group, and the working state of the switch control unit according to the program instructions.

[0057] Through the computer program on the storage medium, flexible viewing angle control, anti-peeping effect and technical effects that adapt to different application scenarios and user needs can be achieved.

[0058] For example, in an in-vehicle display system, a computer program can dynamically adjust the brightness and color of pixel units, as well as the position and shape of the lens group according to the driver's position and line of sight, so that the driver can see all necessary information, while the co-pilot or other passengers may see different content or not be able to see certain content.

[0059] Through the above functions and effects, the storage medium provides a flexible, efficient and customizable control method for the anti-peeping display control device, thereby improving the user experience and the reliability of the device.

[0060] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc. The memory is used to store an operating system, an application program, a boot loader, data, and other programs, such as the program code of the computer program, etc. The memory may also be used to temporarily store data that has been output or is to be output.

[0061] In several embodiments provided in the present application, it is understood that each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved.

[0062] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0063] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A traffic flow prediction method based on block data fusion ARIMA model, characterized in that: S1: Preprocess the collected raw data, and perform data reconstruction and feature extraction on the preprocessed data; S2: combining the feature extraction result with the original data to form multiple data blocks; S3: constructing a block data fusion ARIMA model, and training the block data fusion ARIMA model based on the data block; S4: Perform residual analysis on the trained block data fusion ARIMA model, apply the residual to the random forest algorithm for modeling, capture nonlinear and complex interaction effects, and combine the residual prediction values ​​obtained by the random forest algorithm with the prediction values ​​of the ARIMA model to predict traffic flow.

2. The traffic flow prediction method based on block data fusion ARIMA model according to claim 1 is characterized in that: The step S1 performs data reconstruction and feature extraction on the preprocessed data, specifically including: Reconstruct the time series data into a high-order tensor using multi-path delay embedding transform technology; A high-order tensor is decomposed into the product of multiple low-order tensors, wherein the low-order tensors represent the time trend, periodic changes, and abnormal fluctuations of the data.

3. The traffic flow prediction method based on block data fusion ARIMA model according to claim 2 is characterized in that: The step S3 constructs a block data fusion ARIMA model, specifically including: Integrate the multiple data blocks into the ARIMA model, determine the order q of the MA part through the ACF graph; determine the order p of the AR part through the PACF graph; The order d is determined by using second-order backward difference; Use the AIC criterion to select the optimal ARIMA model parameters; Estimate the coefficients of the AR and MA parts by the MLE method, perform statistical hypothesis tests on these coefficients, and screen for significant coefficients; The model is adjusted according to the significance and refitted until a block data fusion ARIMA model is obtained.

4. The traffic flow prediction method based on block data fusion ARIMA model according to claim 3 is characterized in that: The step S3 further includes training the block data fusion ARIMA model based on the data block: Use statsmodels to import the necessary libraries and use pandas to read the data block data; Train the ARIMA model according to the order q, order p and order d, and calculate MAPE and R 2 Error indicators; Use TimeSeriesSplit to perform cross-validation of time series, evaluate the performance of the block data fusion ARIMA model in different time periods, and retrain the model after finding the optimal parameters.

5. The traffic flow prediction method based on block data fusion ARIMA model according to claim 4 is characterized in that: The residual prediction value obtained by the random forest algorithm is combined with the prediction value of the ARIMA model to predict traffic flow using the following formula: ； in, is the combined prediction result, is a combined prediction model, is the prediction result of the ARIMA model, is the residual partial derivative of the random forest prediction, For time.

6. The traffic flow prediction method based on block data fusion ARIMA model according to claim 5 is characterized in that: The preprocessing of the collected raw data specifically includes: Traffic flow, road conditions, weather and special events are collected as raw data, and the collected raw data are divided into multiple initial data blocks according to time and space. The divided initial data blocks are then subjected to data cleaning, missing value processing and outlier detection preprocessing.

7. A prediction system for a traffic flow prediction method based on block data fusion ARIMA model according to any one of claims 1 to 6, characterized in that: The prediction system comprises: Data collection unit, which collects traffic flow, road conditions, weather and special events as raw data; The first processing unit preprocesses the collected raw data, and performs data reconstruction and feature extraction on the preprocessed data; The second processing unit constructs a block data fusion ARIMA model, trains and optimizes the block data fusion ARIMA model based on the data block, and obtains an optimal block data fusion ARIMA model; The computing unit performs traffic flow prediction based on the optimal block data and the ARIMA model.

8. The prediction system according to claim 7, characterized in that The second processing unit specifically includes: The model building module determines the hyperparameters (p, d, q) required for the ARIMA model by analyzing the characteristics of ACF and PACF, and then selects the optimal model parameters through the AIC value to build the block data fusion ARIMA model; The model optimization module estimates the coefficients of the AR and MA parts through the MLE method, performs statistical hypothesis tests on these coefficients, and screens significant coefficients; adjusts the model according to the significance, and refits until the optimized block data fusion ARIMA model is obtained.

9. A block data fusion ARIMA model, characterized in that: The block data fusion ARIMA model is a block data fusion ARIMA model in a traffic flow prediction method based on a block data fusion ARIMA model as described in any one of claims 1 to 6.

10. A computer-readable storage medium storing computer-executable instructions, characterized in that: When the computer executable instructions are executed by the control processor, a traffic flow prediction method based on a block data fusion ARIMA model as described in any one of claims 1 to 6 is implemented.

Images