CN108492560A - A kind of Road Detection device missing data complementing method and device - Google Patents

Abstract

The invention discloses a kind of Road Detection device missing data complementing methods and device, this method to include：Road Detection device gathered data is analyzed, extraction by the motor vehicle average speed of detector section, each assay intervals by motor vehicle flow and roadway occupancy；It using traffic flow theory as foundation, is fitted using the function of many variables, using speed as dependent variable, flow, occupation rate are that independent variable establishes several function of many variables forms；The function of many variables are fitted based on training set, determine that the most optimized parameter keeps actual value and the root-mean-square error of match value minimum；The evaluation index judged as fitting function quality of root-mean-square error using on verification collection, finally using optimal function and its corresponding the most optimized parameter, missing data progress completion caused by because of the equipment the problems such as.Detector missing data complementing method provided by the invention has easy to implement, easy to operate, the features such as applicability is high, can improve detector gathered data quality, so as to improve road management, promote road traffic efficiency.

Description

A method and device for complementing missing data of a road detector

technical field

The invention relates to expressway traffic information collection, belongs to the technical field of expressway traffic, and in particular relates to a method and device for complementing missing data of a road detector.

Background technique

With the continuous development of the world economy, the transportation industry, especially the road transportation, has also been greatly developed. Especially in recent decades, expressways have developed rapidly around the world due to their characteristics of speed, efficiency, and safety, which have greatly enhanced the economic ties between regions and accelerated the social and economic development of countries and regions in the world.

Traffic data collection is the basis of expressway operation and supervision. Accurate road traffic data supports traffic managers to make effective management decisions.

However, there are various quality problems in the dynamic traffic data collected by the detectors due to aging detection equipment, transmission line failure, poor detection environment, equipment debugging and improper use, etc. How to analyze the existing detector data, and then fill in the missing data accurately and efficiently, has important engineering significance.

Contents of the invention

Purpose of the invention: Aiming at the problems existing in the prior art, the present invention provides a method and device for complementing the missing data of road detectors, complementing the missing data types of road section detectors, and providing traffic managers with a complete expressway traffic data, so as to formulate reasonable management strategies and improve the overall operating efficiency of the expressway network.

Technical solution: In order to achieve the above-mentioned purpose of the invention, the present invention adopts the following technical solution:

A method for complementing missing data of a road detector, comprising the steps of:

(1) Analyze the data collected by the road detector, extract the average speed of motor vehicles passing through the detector section, the flow of motor vehicles passing through each detection interval, and the road occupancy rate;

(2) Specify the type of missing data, and find out the location of each missing data;

(3) Delete the missing data from the original data, and divide the deleted data to form a training set and a verification set;

(4) According to the positive correlation between speed and the ratio of flow and occupancy, add adjustment parameters to establish some multivariate function forms with speed V as dependent variable and flow Q and occupancy O as independent variables for selection and fitting; the established Multivariate functional forms include at least: Among them, β ₁ , β ₂ , β ₃ , β ₄ are adjustment parameters;

(5) In the training set, adjust the undetermined parameters of each multivariate function, and determine the optimal parameters to minimize the root mean square error between the actual value and the fitted value;

(6) In the verification set, use the minimum root mean square error as the basis for selection, calculate the fitting error of each multivariate function, and select the optimal multivariate function form;

(7) Using the optimal function and its corresponding optimization parameters, the independent variable corresponding to the missing data of the detector is substituted into the calculation, and the estimated data is calculated to fill the missing data value.

As a preference, in the step (4), the established multivariate function form also includes at least one of the following:

As preferably, in described step (5), determine the optimal parameter of each multivariate function, comprise: (5.1) set the initial parameter value of multivariate function model; (5.2) utilize least square method principle to establish error function, constantly adjust The parameter value of the model; (5.3) After multiple iterations and adjustments, when the error function value drops to a certain given range, the iteration is stopped, and the model parameters are recorded as the optimal parameters of the multivariate function.

As a preference, in the step (5.2), the method of adjusting the model parameters is as follows: in the training set, assume that the speed is missing data, input the flow Q and the occupancy rate O as independent variables, bring them into the multivariate function to be fitted, and calculate Calculate the estimated velocity V', use the error function as the basis for judgment, and continuously adjust the parameter values of the model in the direction that makes the error function value decrease the fastest, so that the error between the estimated velocity V' and the real velocity V gradually decreases.

As a preference, in the step (6), the method for selecting the optimal multivariate function is: applying the optimization parameters of each multivariate function, substituting the independent variable in the verification set data into the multivariate function, and calculating the estimated value of the missing data type, Then calculate the root mean square error of each multivariate function in the verification set to fit the estimated value and the measured value, compare the various multivariate functions, and select the function form corresponding to the smallest root mean square error.

Another aspect of the present invention provides a road detector missing data complement device, comprising:

The data extraction module is used to analyze the data collected by the road detector, and extract the average speed of motor vehicles passing through the detector section, the flow of motor vehicles passing through each detection interval, and the road occupancy rate;

The missing location module is used to clarify the type of missing data and find out the location of each missing data;

The preprocessing module is used to delete missing data from the original data, and divide the deleted data to form a training set and a verification set;

The multivariate function building block is used to add adjustment parameters to establish a number of multivariate function forms with speed V as the dependent variable and flow Q and occupancy O as independent variables according to the positive correlation between the speed and the ratio of flow and occupancy. ; The established multivariate function forms include at least: Among them, β ₁ , β ₂ , β ₃ , β ₄ are adjustment parameters;

The function fitting module is used to adjust the undetermined parameters of each multivariate function based on the training set, and determine the optimal parameters to minimize the root mean square error between the actual value and the fitted value;

The optimal function selection module is used to calculate the fitting error of each multivariate function based on the verification set with the minimum root mean square error as the selection basis, and select the optimal multivariate function form;

And, the missing filling module is used to apply the optimal function and its corresponding optimization parameters, substitute the independent variable corresponding to the missing data of the detector for calculation, and calculate estimated data to fill the missing data value.

Preferably, the multivariate function forms established in the multivariate function building block also include at least one of the following:

Beneficial effect: compared with prior art, the beneficial effect of the present invention is:

1, the multivariate function fitting form that the present invention provides meets the ratio of speed and flow and density in the traffic flow theory and is positively correlated, and density and time occupancy are positively correlated, and the ratio of speed and flow and occupancy is positively correlated with each other, thus Existing traffic and occupancy data can be used to estimate missing speed data. 2. The present invention uses the root mean square error as the evaluation standard, and selects the multivariate function fitting form with the lowest error for each missing data set, which can ensure higher completion accuracy. 3. Complementing the missing speed data of roads in the present invention can improve the traffic operation information of expressways, provide a basis for traffic managers to formulate management strategies, and then improve the safety of traffic operation and improve the overall operation level of expressways.

Description of drawings

Fig. 1 is the flow chart of complementing the missing data of the road detector of the present invention;

Fig. 2 is a partial sample screenshot of data collected by the road detector of the present invention;

Fig. 3 is a diagram of error analysis results of multivariate function for missing data completion in the example of the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

As shown in Figure 1, a method for complementing missing data of a road detector disclosed in an embodiment of the present invention firstly analyzes the collected data and extracts useful information therein; defines the missing data and deletes the missing data from the original data, And the deleted data is divided to form a training set and a verification set. Then establish the multivariate function form for selection and fitting, adjust the undetermined parameters of each multivariate function, and determine the optimal parameters to minimize the root mean square error between the actual value and the fitting value. Then, the root mean square error is used as the basis for selection, and the fitting error of each multivariate function is calculated to select the optimal form of multivariate function. Finally, the optimal function is substituted into the calculation to fill in the missing data values. The detailed steps are as follows:

(1) Analyze the data collected by the road detector, and extract the average speed of motor vehicles passing through the detector section, the flow of motor vehicles passing through each detection interval, and the road occupancy rate. First, collect the data collected by road detectors (mainly coil detectors and geomagnetic detectors) on the road, and then conduct preliminary filtering and analysis on the collected traffic collection information to collect useful information, and the traffic collected by road section detectors is effective. The information mainly includes traffic flow, speed, and road occupancy rate. The collection interval is related to the type of detector, generally between 30 seconds and 300 seconds.

(2) Identify the type of missing data, and find out the location of each missing data. In this step, the collected data is analyzed to clarify the type of missing detector data. In Figure 2, it can be seen that some speed values detected by the detector are missing or have abnormal data. Therefore, in this example, the average speed of the traffic flow is used as the data type to be filled in to find the location of the abnormal speed data.

(3) Delete the missing data from the original data, and divide the deleted data to form a training set and a verification set. In this step, the data rows corresponding to the missing data in the data set are deleted, and the data in each data set sample is roughly divided into two sections from the middle. Validation collection. In this example, the data rows with missing or abnormal speeds are deleted and divided into two, one part is used as a training set and the other part is used as a validation set.

(4) Establish a multivariate function form for selection and fitting. According to the traffic flow theory, the speed is proportional to the ratio of flow and density, and the traffic density is proportional to the time occupancy, so the speed is positively correlated with the ratio of flow and occupancy. In this example, the existing accurate data of the detector are the flow rate Q and the occupancy rate O, and the missing data to be supplemented is the velocity parameter V. Therefore, with V (velocity) as the dependent variable, Q (flow rate) and O (occupancy rate) as independent variables, and βi as undetermined parameters, 7 multivariate function forms are established as fitting functions. The seven fitting functions are:

first functional form

The second functional form

The third functional form

The fourth functional form

fifth functional form

sixth functional form

The seventh function form

The above 7 function forms have been tested by a large number of samples and have a high degree of matching. Among them, the fourth, fifth and sixth function forms are the most universal. In actual operation, all or part of the functions can be selected for fitting and screening .

(5) In the training set, adjust the undetermined parameters for each multivariate function to determine the optimal parameters. In this step, each multivariate function fitting mainly includes the following steps:

Step (5.1) sets initial parameter values.

Step (5.2) uses the principle of the least square method to establish an error function, and continuously adjusts the parameter values of the model.

After multiple iterations and adjustments in step (5.3), when the value of the error function drops to a certain given range, the iteration is stopped, and the model parameters are recorded as the optimization parameters of the multivariate function.

The optimal undetermined parameters of each of the seven fitting functions are finally determined by the above steps.

Specifically, first set the initial value of the undetermined parameter βi in each function, and bring the flow Q and occupancy O of each detector data in the training set into the seven fitting functions as independent variables to obtain the estimated speed V', using the least squares principle to establish an error function as a criterion for parameter optimization, continuously adjust the parameter values of the model in the direction that makes the error function value drop the fastest, and finally find the estimated speed V' and the real speed V in each training set The optimal undetermined parameter βi value with the smallest average error.

(6) Using the minimum root mean square error as the selection basis, calculate the fitting error of each multivariate function, and select the optimal multivariate function form. In this step, the root mean square error is used as the basis for selecting the multivariate function, and the root mean square error is the arithmetic square root of the mean square error of the selected data, also known as the standard error. The basic formula is as follows, where observed _t represents the actual detected value, predicted _t represents the calculated estimated value, and N is the total number of samples.

Take out the verification set, input the flow Q and occupancy rate O collected by each detector in the verification set as independent variables, bring them into each fitting function, and calculate the estimated velocity V' by using its optimal undetermined parameters. Calculate the root mean square error between the estimated velocity V' and the true velocity V in the validation set and record it.

Compare the root mean square error between the estimated speed V' calculated by the seven fitting functions and the real speed V, and select the fitting function corresponding to the smallest error as the missing completion function of the data. The optimal function corresponding to this function is to be determined. Parameter βi is the optimization parameter.

(7) Using the optimal function and its corresponding optimization parameters, the independent variable corresponding to the missing data of the detector is substituted into the calculation, and the estimated data is calculated to fill the missing data value. In this example, the flow Q and occupancy rate O collected by the detector corresponding to the missing speed data are used as independent variables to input the function, and the estimated speed value is obtained, so as to fill in the deleted missing speed value. It can be seen from Figure 3 that the root mean square error calculated by the fourth multivariate function form of the selected data has a minimum error value of 8.21%. Using this function as the missing completion function of the data, the final filling accuracy is above 90%.

Another embodiment of the present invention discloses a missing data complementing device for a road detector, comprising: a data extraction module for analyzing the data collected by the road detector, extracting the average speed of the motor vehicle passing through the detector section, the passage of each detection interval The motor vehicle flow and road occupancy rate; the missing location module is used to clarify the type of missing data, and find out the location of each missing data; the preprocessing module is used to delete the missing data from the original data, and The deleted data is divided to form a training set and a verification set;

The multivariate function building block is used to add adjustment parameters to establish a number of multivariate function forms with speed V as the dependent variable and flow Q and occupancy O as independent variables according to the positive correlation between the speed and the ratio of flow and occupancy. ; The established multivariate function forms include at least: The function fitting module is used to adjust the undetermined parameters of each multivariate function based on the training set, and determine the optimal parameters to minimize the root mean square error between the actual value and the fitted value; the optimal function selection module is used to use the mean square value based on the verification set The minimum root error is used as the basis for selection, and the fitting error of each multivariate function is calculated to select the optimal multivariate function form; and, the missing filling module is used to apply the optimal function and its corresponding optimization parameters to correspond to The independent variable is substituted into the calculation, and the estimated data is calculated to fill in the missing data values.

The above embodiment of the missing data complementing device for road detectors can be used to implement the above embodiment of the missing data complementing method for road detectors. The technical principles, technical problems solved and technical effects produced are similar. For the specific working process and relevant descriptions of the data completion, reference may be made to the corresponding process in the above-mentioned embodiment of the missing data completion method of the road detector, and details are not repeated here.

Those skilled in the art can understand that the modules in the embodiment can be adaptively changed and installed in one or more systems different from the embodiment. Modules or units or components in the embodiments may be combined into one module or unit or component, and furthermore may be divided into a plurality of sub-modules or sub-units or sub-assemblies.

The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements can also be made, and these improvements should also be regarded as the present invention. scope of protection.

Claims (7)

1. A road detector missing data complement method is characterized in that: the method comprises the steps: (1) Analyze the data collected by the road detector, extract the average speed of motor vehicles passing through the detector section, the flow of motor vehicles passing through each detection interval, and the road occupancy rate; (2) Specify the type of missing data, and find out the location of each missing data; (3) Delete the missing data from the original data, and divide the deleted data to form a training set and a verification set; (4) According to the positive correlation between speed and the ratio of flow and occupancy, add adjustment parameters to establish some multivariate function forms with speed V as dependent variable and flow Q and occupancy O as independent variables for selection and fitting; the established Multivariate functional forms include at least: Among them, β ₁ , β ₂ , β ₃ , β ₄ are adjustment parameters; (5) In the training set, adjust the undetermined parameters of each multivariate function, and determine the optimal parameters to minimize the root mean square error between the actual value and the fitted value; (6) In the verification set, use the minimum root mean square error as the basis for selection, calculate the fitting error of each multivariate function, and select the optimal multivariate function form; (7) Using the optimal function and its corresponding optimization parameters, the independent variable corresponding to the missing data of the detector is substituted into the calculation, and the estimated data is calculated to fill the missing data value. 2. A kind of road detector missing data completion method according to claim 1, is characterized in that: in described step (4), the multivariate function form that establishes also comprises following at least one kind: 3. A kind of road detector missing data complement method according to claim 1, is characterized in that: in described step (5), determine the optimization parameter of each multivariate function, comprise: (5.1) set multivariate The initial parameter value of the function model; (5.2) Use the least square method to establish the error function, and continuously adjust the parameter value of the model; (5.3) After multiple iterations and adjustments, when the error function value drops to a given range, Stop iterating and record the model parameters as the optimization parameters for this multivariate function. 4. a kind of road detector missing data completion method according to claim 3, is characterized in that: in described step (5.2), the method for adjusting model parameter is: in training set, suppose speed is missing data, will The flow Q and occupancy O are input as independent variables and brought into the multivariate function to be fitted to calculate the estimated velocity V', and the error function is used as the basis for judgment, and the parameter values of the model are continuously adjusted in the direction that makes the error function value decrease the fastest , so that the error between the estimated velocity V' and the real velocity V gradually decreases. 5. a kind of road detector missing data supplementary method according to claim 1, is characterized in that: in described step (6), the method for selecting optimal multivariate function is: apply the optimization parameter of each multivariate function, Substitute the independent variables in the verification set data into the multivariate function, calculate the estimated value of the missing data type, and then calculate the root mean square error between the estimated value and the measured value of each multivariate function in the verification set, compare each multivariate function, and select The functional form corresponding to the smallest root mean square error. 6. A road detector missing data complement device, characterized in that, comprising: The data extraction module is used to analyze the data collected by the road detector, and extract the average speed of motor vehicles passing through the detector section, the flow of motor vehicles passing through each detection interval, and the road occupancy rate; The missing location module is used to clarify the type of missing data and find out the location of each missing data; The preprocessing module is used to delete missing data from the original data, and divide the deleted data to form a training set and a verification set; The multivariate function building block is used to add adjustment parameters to establish a number of multivariate function forms with speed V as the dependent variable and flow Q and occupancy O as independent variables according to the positive correlation between the speed and the ratio of flow and occupancy. ; The established multivariate function forms include at least: Among them, β ₁ , β ₂ , β ₃ , β ₄ are adjustment parameters; The function fitting module is used to adjust the undetermined parameters of each multivariate function based on the training set, and determine the optimal parameters to minimize the root mean square error between the actual value and the fitted value; The optimal function selection module is used to calculate the fitting error of each multivariate function based on the verification set with the minimum root mean square error as the selection basis, and select the optimal multivariate function form; And, the missing filling module is used to apply the optimal function and its corresponding optimization parameters, substitute the independent variable corresponding to the missing data of the detector for calculation, and calculate estimated data to fill the missing data value. 7. A kind of road detector missing data completion device according to claim 6, is characterized in that, the multivariate function form established in the described multivariate function construction module also includes following at least one: