Disclosure of Invention
The invention aims to provide a vehicle following optimizing control method based on fuzzy reasoning, which solves the problem that vehicle following research in the prior art is easy to be interfered by noise.
The invention discloses a vehicle following optimizing control method based on fuzzy reasoning, which is characterized by comprising the following steps:
step 1, screening road data of vehicle driving to obtain vehicle following data in vehicle track data;
step 2, calculating the distance between the vehicle following data obtained in the step 1 through a dispersion square sum formula, taking the calculation result as the similarity of a condensed hierarchical clustering algorithm, continuously combining the vehicle following data based on the minimum similarity, and finally dividing the vehicle following data into a plurality of clusters; combining clustered data into a cluster class through a hierarchical clustering algorithm to form a tree graph hierarchical structure, setting a similarity threshold value theta as a clustering division basis of vehicle following data, and enabling a clustering result of the vehicle following data division to be optimal;
step 3, obtaining a clustering and dividing result of the vehicle following data by utilizing the step 2, defining a fuzzy rule, enabling divided clusters to be input fuzzy sets, obtaining a central point of each cluster by adding and averaging vehicle following data in each divided cluster, calculating membership degree of input quantity of each fuzzy set by utilizing an FCM algorithm, and constructing E (l) Measuring the value of the data membership degree and further setting the confidence degree bel (l) ;
Step 4, firstly, calculating a weight multiplied by the membership degree by utilizing the fuzzy set defined in the step 3 and the calculated membership degree; then, improving the central value of the fuzzy set by taking the clustering cluster as the input quantity by utilizing the trust degree set in the step 3; thirdly, calculating the confidence coefficient of the rule; and finally, a central average deblurring method is used, so that the efficiency and the reliability of the optimizing solution are improved.
The present invention is also characterized in that,
the step 1 is specifically implemented according to the following steps:
according to the key fields in the road data of the selected vehicle, finding out the following data of the vehicle, and further screening the following data of the vehicle according to the custom rules, wherein the custom rules are as follows: only the following behaviors among vehicles of the same type are researched, data of vehicles changing lanes in the driving process and data of vehicles on multiple passenger lanes are removed, and data with the following time exceeding 45s are reserved so as to ensure the integrity of the following process, so that vehicle following data are obtained.
The step 2 is specifically implemented according to the following steps:
step 2.1, calculating the distance:
the degree of dispersion between the vehicle following data is represented by the sum of squares of dispersion, sum of squares of dispersion S r The formula is as follows:
in (x) i For the screening processing of the vehicle following data in the r cluster,the gravity center of the cluster obtained by clustering the vehicle following data is n r Is the number of cluster data, r epsilon n r I is the index of the ith cluster class, i e n r ;
Let d (c) r ,c s ) Is cluster c obtained after clustering r And adjacent cluster class c s When cluster c r And adjacent cluster class c r The distance is relatively short, the cluster distance can be calculated through the sum of squares of the deviations, the calculation result is used as the similarity of the condensed hierarchical clustering algorithm, the vehicle following data is combined according to the principle of minimum similarity, and a cluster distance calculation formula d (c) r ,c s ) The following are provided:
d(c r ,c s )=(S w -S r -S s ) 1/2
s in w Representing cluster c r And adjacent cluster class c s Sum of squares of dispersion of the merged cluster, S r Representing cluster c r Sum of squares of dispersion of S s Representing adjacent cluster class c s The sum of the squares of the deviations of (a);
step 2.2, hierarchical clustering:
step 2.2.1, regarding each vehicle following data as a unit element cluster during initial calculation, and calculating the distance between the vehicle following data by using the distanceThe formula calculates, and the calculation result is used as the similarity of the condensed hierarchical clustering algorithm and is stored in a similarity matrix D= [ D (c) i ,c j )]In (a) and (b);
wherein c i C, a unit pixel cluster with the serial number i in the following data of the vehicle j The adjacent single element cluster with the sequence number j in the vehicle following data is used as a cluster class;
setting the level L (m) of the preliminary cluster to 0, i.e., L (m) =0;
wherein m represents the number of layers of the cluster;
step 2.2.2, finding out two single-element clusters with the nearest similarity in the vehicle following data by comparing the distance values stored in the similarity matrix in step 2.2.1;
step 2.2.3, accumulating the clustering layers of the vehicle following data, merging the two single element cluster classes with the nearest similarity found in the step 2.2.2, thereby generating a new multi-element cluster, and determining the clustering cluster class c of the layer at the moment r And adjacent cluster class c s Position D [ c ] in the matrix r ,c s ]So that the hierarchy of the current cluster L (m) =d [ c ] r ,c s ];
Wherein (c) r ,c s ) A new cluster number is generated; if the initial single-element cluster class in the similarity matrix is replaced completely, merging the multi-element cluster classes;
step 2.2.4, updating the similarity matrix D, and deleting the cluster class c r And adjacent cluster class c s Adding the value of the corresponding position of the newly generated cluster in the similarity matrix D at the corresponding position of the matrix;
step 2.2.5, if the termination condition is met, stopping clustering, otherwise, turning to step 2.2.2;
wherein, the termination condition is: all vehicle-following data are combined into one cluster;
step 2.3, hierarchical clustering is carried out on the vehicle following data through the step 2.2 to construct a tree diagram hierarchical structure, wherein the horizontal axis in the tree diagram hierarchical structure represents the serial number of the data, the vertical axis represents the distance between the data, and the original data is positioned at the lowest layer of the tree;
and 2.4, setting the threshold value theta to be 0.2 times of the maximum similarity by utilizing the Balaiduo law, and ensuring that the number of clustering of vehicle following data is proper so as to reduce the operation time and have higher reliability.
The step 3 is specifically implemented according to the following steps:
step 3.1, defining a fuzzy rule:
clustering the vehicle following data in the step 2 to form cluster class, and taking the cluster class as n-input single-output sample data T= { (x) (p) ;y (p) ) P=1, 2, …, N, one rule for each sample data;
wherein N is the number of divided cluster types, N is the number of input data in a group of cluster types, and x (p) Inputting data for the p-th clustery (p) Representing corresponding output data;
the extracted IF-THEN rule is expressed as:
IF x i1 isand...and x im is/>and THEM y is B (l)
wherein,cluster class clustered for vehicle following data is used as input quantity fuzzy set, B (l) Fuzzy set of output y on rule number l, l=1, 2,..m represents the number of rules, M represents the total number of rules, (i 1..im) is a subsequence of (1..n), and m.ltoreq.n;
step 3.2, calculating membership degree by using FCM algorithm:
calculating membership mu by using FCM algorithm ij (l) The formula is as follows:
wherein d ij Representing vehicle following data x j Distance d from central point of cluster i after hierarchical clustering kj Representing vehicle following data x j Distance x from center point of cluster k after hierarchical clustering j N is the number of clusters after hierarchical clustering, i is the cluster after the i-th hierarchical clustering, k is the cluster after the k-th hierarchical clustering, j is the index of the vehicle following data, and l is the index of the current rule sequence number;
dividing the vehicle following data into corresponding cluster types by hierarchical clustering, calculating the membership degree of the vehicle following data to the center distance of each divided cluster type, ifIndicating the risk of collision if +.>Indicating that the rear-end collision is separated from the following state, and indicating that the vehicle is in a safe and stable running environment at present;
step 3.3, calculating the measurement value:
set E (l) Measuring vehicle following data x j Membership mu of (C) ij According to the definition of entropy, calculating a value formula E of membership degree of measurement and relaxation data (l) The following are provided:
wherein c is the number of clusters after hierarchical clustering, and x j For the vehicle following data,for vehicle following data x j I is the index of the ith cluster, j is the index of the following data of the vehicle;
step 3.4, setting trust:
utilization stepMeasuring membership value E in step 3.3 (l) Setting vehicle following data x j Is bej (l) Trust bej (l) The calculation formula is as follows:
where N represents the total number of vehicle-following data samples.
Step 4 is specifically implemented according to the following steps:
step 4.1 if the termination condition is satisfiedNo rule is generated and the method terminates; otherwise, the fuzzy set defined in the step 3 and the calculated membership are utilized to calculate the weight multiplied by the membership, and the formula is as follows:
wherein A is ij Obtaining cluster class for hierarchical clustering of the vehicle following data, wherein each cluster class corresponds to a fuzzy set of input quantity,representative heel data->Inputting membership value of the fuzzy set at p;
each cluster obtained after hierarchical clustering of the vehicle following data is further replaced in step 2.2.4 by calculating membership to obtain a weight multiplied by the membership, a similarity matrix is updated, and the similarity of the newly generated cluster and other clusters in the similarity matrix is calculated to further accelerate the hierarchical clustering speed;
step 4.2, aiming at the rules defined in the step 3, taking the cluster class of the vehicle following datamation partition as an input fuzzy set, wherein each input quantity corresponds to one rule, generating fuzzy rules, and if the same repeated rules of the previous parts exist in the rules, merging the extracted rules;
set the central value in each rule sequence number l asImproving the central value by using the confidence level of the input fuzzy set samples and the confidence level among the input fuzzy set samples, wherein the improved central value is +.>The calculation formula is as follows:
in doc (i) To input fuzzy set sample confidence, bel (l) The trust degree among input fuzzy set samples is that N is the number of the input fuzzy set samples by taking the cluster differentiated by the vehicle following data as the current rule sequence number index;
step 4.3, generating a fuzzy rule as follows:
IF x i1 isand…and x im is/>THEN y is B (l)
wherein B is (l) Is a Gaussian distribution membership functionIs a fuzzy set of Gaussian distribution membership function +.>Medium parameter sigma (l) The calculation is as follows:
wherein y is (l) For the rule center value of rule number l,the central value of the sequence number of the input fuzzy set with the rule sequence number of l, u (l) Obtaining m membership products of cluster class for hierarchical clustering of vehicle following data, and performing +.>Membership product of sequence number of input fuzzy set with rule sequence number of l +.>An index corresponding to the kth data of the output fuzzy aggregation data with the rule sequence number of l is represented; a is that ij Obtaining cluster class for hierarchical clustering of the vehicle following data, wherein m represents the number of output fuzzy sets of rule l;
calculating confidence doc of rule (l) The following are provided:
in the method, in the process of the invention,center value of sequence number of output fuzzy set for rule sequence number l, < >>For regular sequence numbersThe central value of the sequence number of the input fuzzy set is l, k is the kth output model, and t is the t input model;
calculating the confidence coefficient of the rule to evaluate the rationality degree of the rule, and indicating that no conflict exists between the merged rules;
step 4.4, obtaining a prediction model by using a central average deblurring method;
let the activation degree of rule number l be act (l), the formula of activation degree act (l) is as follows:
in the method, in the process of the invention,fuzzy set A with rule sequence number of l ij Membership degree of (3);
let each rule center point and the prediction point construct a prediction formula f (x), the prediction formula f (x) is as follows:
in the method, in the process of the invention,is a rule center with a rule sequence number of l, and M is the rule number;
by calculating the central value of each ruleDescribing rule center point +.>Calculating each new rule generated by the relation with the current vehicle following data x' by using a prediction formula f (x) to finally obtain a predicted following data result of the following vehicle at the next moment, and enabling a driver to obtain the data of the following vehicle at the next moment according to the predictionAnd adjusting the current running speed, if the vehicle rear-end collision is predicted, immediately decelerating the vehicle by the driver, and if the predicted data are always in a safe and stable state, keeping the vehicle to be in a following state all the time by the driver.
The vehicle following optimizing control method based on fuzzy reasoning has the beneficial effects that fuzzy interval division is firstly carried out on input and output variables. Compared with the WM algorithm which needs to artificially set the fuzzy intervals of the input and output variables, the method has the advantages that the generated fuzzy rule base is not complete because the interval division intuitively influences the extraction of the fuzzy rules. Aiming at the lack of good robustness of the WM algorithm, the invention introduces information entropy and trust degree to weight the data. When noise data exists in vehicle following, rules with lower reliability and even errors are easily extracted. Therefore, the invention screens the data according to the following characteristics of the vehicle and the response time of the driver during the data preprocessing. And analyzing the following behavior of the driver to determine the input and output variables of the following model. Because of the lack of higher robustness of the WM method, the trust level formula in the HC_WM algorithm is provided to express the trust level value of the sample belonging to the real data set. In addition, in the process of extracting the fuzzy rule, the trust degree is added into a calculation formula of the output quantity of the rule center point, so that the robustness of the WM method is improved.
Detailed Description
The invention will be described in detail below with reference to the drawings and the detailed description.
The problems of the traditional WM method are improved and optimized. The traditional WM method needs to directly extract fuzzy rules from samples through the following 3 steps.
Step 1, rule extraction
Step 11, set input amount x= (x) 1 ,...x n ) And input quantity x i Is divided into m i The complete fuzzy rule numbers share the same fuzzy setA strip.
Step 1.2, data-rule conversion
Step 1.2.1, p=1, data (x (p) ,y (p) ) Is set according to the input quantity of each of the input valuesConversion to fuzzy set->And calculate at fuzzy set +.>The corresponding membership degree in the formula is:
step 1.2.2, per input quantityMay be mapped to a plurality of fuzzy sets, and the fuzzy set with the largest membership degree is selectedAs input quantity +.>Is defined by a fuzzy set of data (x (p) ,y (p) ) The extracted fuzzy rule is as follows:
if x 1 isand...and x n is/>Then y is y (i)
step 1.2.3, calculation data (x (p) ,y (p) ) The weight of the corresponding fuzzy rule is as follows:
step 1.2.4, if i=k, completing rule extraction of all data, and ending the extraction rule; otherwise, i=i+1, jump to step 1.2.2 and continue extracting rules.
Step 2, rule merging
Step 2.1, the fuzzy rules with the same front piece are divided into a group, M groups are shared, and the first group comprises N l A bar rule; and combining the values of the regular back part y to obtain a set B l 。
Step 2.2, utilizing triangular membership functionBuilding corresponding membership degree on discourse domain, and calculating yc l The center point, the formula is as follows:
step 2.3, calculating sigma l The formula is as follows:
step 2.4, after merging the rules, each group corresponds to a fuzzy rule, and the fuzzy rule is as follows:
if x 1 isand...and x n is/>then y is B l .
calculating confidence doc of the combined rule l The formula is as follows:
step 3, rule speculation
Step 3.1, find the neighbor with the largest number in the extrapolation rule set, and call this set max-group. Each rule in max-group is calculated as follows:
b is the neighbor number owned by the rule to be pushed from the rule generated by the data; l (L) r Is the index number of the adjacent rule; dis (dis) r Is the distance between the input center of the rule and its neighboring rule.
Step 3.2, generating by extrapolation rules of max-group:
if x 1 isand...and x n is/>Then y is B l .
wherein B is l Is a triangle fuzzy set with membership function ofWherein sigma l The specific calculation formula is shown in (2-15):
and 3.3, calculating the credibility of an extrapolation rule, wherein the formula is as follows:
and 3.3, repeating the step 3.1, and finding out the next extrapolation rule.
Although the above method can be improved by using the inference method of unknown rules by using adjacent known rules, the specific process is shown in fig. 2, the number of inferred rules still cannot meet the actual situation requirement, and the rule base has the problem of imperfection. In addition, the conventional WM method uses a single sample data extraction rule, and once the sample data is noisy, the WM algorithm is easily disturbed by erroneous sample data, thus lacking in high integrity and robustness. Therefore, in order to improve the integrity of the WM method, the invention uses all sample data to calculate the output variables of the current rule and enhances the MW method based on hierarchical clustering.
The invention discloses a vehicle following optimizing control method based on fuzzy reasoning, which is implemented according to the following steps:
as shown in fig. 3 and fig. 4, step 1, screening road data of vehicle running to obtain vehicle following data in vehicle track data;
the step 1 is specifically implemented according to the following steps:
according to the key fields in the road data of the selected vehicle, finding out the following data of the vehicle, and further screening the following data of the vehicle according to the custom rules, wherein the custom rules are as follows: only the following behaviors among vehicles of the same type are researched, data of vehicles changing lanes in the driving process and data of vehicles on multiple passenger lanes are removed, and data with the following time exceeding 45s are reserved so as to ensure the integrity of the following process, so that vehicle following data are obtained.
Step 2, calculating the distance between the vehicle following data obtained in the step 1 through a dispersion square sum formula, taking the calculation result as the similarity of a condensed hierarchical clustering algorithm, continuously combining the vehicle following data based on the minimum similarity, and finally dividing the vehicle following data into a plurality of clusters; combining clustered data into a cluster class through a hierarchical clustering algorithm to form a tree graph hierarchical structure, setting a similarity threshold value theta as a clustering division basis of vehicle following data, and enabling a clustering result of the vehicle following data division to be optimal;
the step 2 is specifically implemented according to the following steps:
step 2.1, calculating the distance:
the degree of dispersion between the vehicle following data is represented by the sum of squares of dispersion, sum of squares of dispersion S r The formula is as follows:
in (x) i For the screening processing of the vehicle following data in the r cluster,the gravity center of the cluster obtained by clustering the vehicle following data is n r Is the number of cluster data, r epsilon n r I is the index of the ith cluster class, i e n r ;
Let d (c) r ,c s ) Is cluster c obtained after clustering r And adjacent cluster class c s When cluster c r And adjacent cluster class c r The distance is relatively short, the cluster distance can be calculated through the sum of squares of the deviations, the calculation result is used as the similarity of the condensed hierarchical clustering algorithm, the vehicle following data is combined according to the principle of minimum similarity, and a cluster distance calculation formula d (c) r ,c s ) The following are provided:
d(c r ,c s )=(S w -S r -S s ) 1/2
s in w Representing cluster c r And adjacent cluster class c s Sum of squares of dispersion of the merged cluster, S r Representing cluster c r Sum of squares of dispersion of S s Representing adjacent cluster class c s The sum of the squares of the deviations of (a);
step 2.2, hierarchical clustering:
step 2.2.1, in the initial calculation, regarding each vehicle following data as a unit pixel cluster, calculating the distance between the vehicle following data by using a distance calculation formula, and storing the calculation result as the similarity of the condensed hierarchical clustering algorithm in a similarity matrix D= [ D (c) i ,c j )]In (a) and (b);
wherein c i C, a unit pixel cluster with the serial number i in the following data of the vehicle j The adjacent single element cluster with the sequence number j in the vehicle following data is used as a cluster class;
setting the level L (m) of the preliminary cluster to 0, i.e., L (m) =0;
wherein m represents the number of layers of the cluster;
step 2.2.2, finding out two single-element clusters with the nearest similarity in the vehicle following data by comparing the distance values stored in the similarity matrix in step 2.2.1;
step 2.2.3, accumulating the clustering layers of the vehicle following data, merging the two single element cluster classes with the nearest similarity found in the step 2.2.2, thereby generating a new multi-element cluster, and determining the clustering cluster class c of the layer at the moment r And adjacent cluster class c s Position D [ c ] in the matrix r ,c s ]So that the hierarchy of the current cluster L (m) =d [ c ] r ,c s ];
Wherein (c) r ,c s ) A new cluster number is generated; if the initial single-element cluster class in the similarity matrix is replaced completely, merging the multi-element cluster classes;
step 2.2.4, updating the similarity matrix D, and deleting the cluster class c r And adjacent cluster class c s Adding the value of the corresponding position of the newly generated cluster in the similarity matrix D at the corresponding position of the matrix;
step 2.2.5, if the termination condition is met, stopping clustering, otherwise, turning to step 2.2.2;
wherein, the termination condition is: all vehicle-following data are combined into one cluster;
step 2.3, hierarchical clustering is carried out on the vehicle following data through the step 2.2 to construct a tree diagram hierarchical structure, wherein the horizontal axis in the tree diagram hierarchical structure represents the serial number of the data, the vertical axis represents the distance between the data, and the original data is positioned at the lowest layer of the tree;
and 2.4, setting the threshold value theta to be 0.2 times of the maximum similarity by utilizing the Balaiduo law, and ensuring that the number of clustering of vehicle following data is proper so as to reduce the operation time and have higher reliability.
Step 3, obtaining a clustering and dividing result of the vehicle following data by utilizing the step 2, defining a fuzzy rule, enabling divided clusters to be input fuzzy sets, obtaining a central point of each cluster by adding and averaging vehicle following data in each divided cluster, calculating membership degree of input quantity of each fuzzy set by utilizing an FCM algorithm, and constructing E (l) Measuring the value of the data membership degree and further setting the confidence degree bel (l) ;
The step 3 is specifically implemented according to the following steps:
step 3.1, defining a fuzzy rule:
clustering the vehicle following data in the step 2 to form cluster class, and taking the cluster class as n-input single-output sample data T= { (x) (p) ;y (p) ) P=1, 2, …, N, one rule for each sample data;
wherein N is the number of divided cluster types, N is the number of input data in a group of cluster types, and x (p) Inputting data for the p-th clustery (p) Representing corresponding output data;
the extracted IF-THEN rule is expressed as:
IF x i1 isand…and x im is/>and THEM y is B (l)
wherein,cluster class clustered for vehicle following data is used as input quantity fuzzy set, B (l) Fuzzy set of output y on rule number l, l=1, 2,..m represents the number of rules, M represents the total number of rules, (i 1, … im) is a subsequence of (1, … n), and m.ltoreq.n;
in the invention, m=n is selected, namely, all the following data in the cluster class divided by the vehicle following data cluster are used as the precondition of the extracted IF-THEN rule, so that the integrity of the WM method is improved.
Step 3.2, calculating membership degree by using FCM algorithm:
calculating membership mu by using FCM algorithm ij (l) The formula is as follows:
wherein d ij Representing vehicle following data x j Distance d from central point of cluster i after hierarchical clustering kj Representing vehicle following data x j Distance x from center point of cluster k after hierarchical clustering j N is the number of clusters after hierarchical clustering, i is the cluster after the i-th hierarchical clustering, k is the cluster after the k-th hierarchical clustering, j is the index of the vehicle following data, and l is the index of the current rule sequence number;
dividing the vehicle following data into corresponding cluster types by hierarchical clustering, calculating the membership degree of the vehicle following data to the center distance of each divided cluster type, and if mu ij (l) =1, indicating the risk of collision if μ ij (l) When the number of fuzzy sets is less than 0, the tail end collision following state is separated, so that the fact that the tail end collision following state is in a safe and stable driving environment at present is indicated, the number of fuzzy sets of input output quantity can greatly influence the accuracy of the extraction rule by adopting FCM to calculate the membership degree;
step 3.3, calculating the measurement value:
set E (l) Measuring vehicle following data x j Membership mu of (C) ij According to the definition of entropy, calculating a value formula E of membership degree of measurement and relaxation data (l) The following are provided:
wherein c is the number of clusters after hierarchical clustering, and x j For the vehicle following data,for vehicle following data x j I is the index of the ith cluster, j is the index of the following data of the vehicle;
step 3.4, setting trust:
using the measure of membership value E in step 3.3 (l) Setting vehicle following data x j Is bej (l) Trust bej (l) The calculation formula is as follows:
wherein N represents the total number of the vehicle following data samples;
when the WM algorithm extracts the vehicle following data, the confidence level is set for the vehicle following sample data because of the problem that the extraction process is disturbed due to noise. When trust bej is performed on sample data with noise (l) Bej when calculating (l) The larger the value, the higher the credibility of the noise data is, according to the characteristic, bej is eliminated (l) And the data with large values reduces noise data in the data samples, so that a good rule set can be obtained during rule extraction, and the robustness of the WM algorithm is enhanced.
The reliability of rear-end collision or safety state in the process of predicting the vehicle running through the membership is enhanced by improving the membership and setting corresponding trust to weight the vehicle following data. When the WM algorithm extracts vehicle following data, the problem that the extraction process is interfered due to noise is solved, and the robustness of the WM algorithm is enhanced.
Step 4, firstly, calculating a weight multiplied by the membership degree by utilizing the fuzzy set defined in the step 3 and the calculated membership degree; then, improving the central value of the fuzzy set by taking the clustering cluster as the input quantity by utilizing the trust degree set in the step 3; thirdly, calculating the confidence coefficient of the rule; and finally, a central average deblurring method is used, so that the efficiency and the reliability of the optimizing solution are improved.
Step 4 is specifically implemented according to the following steps:
step 4.1 if the termination condition is satisfiedNo rule is generated and the method terminates; otherwise, the fuzzy set defined in the step 3 and the calculated membership are utilized to calculate the weight multiplied by the membership, and the formula is as follows:
wherein A is ij Obtaining cluster class for hierarchical clustering of the vehicle following data, wherein each cluster class corresponds to a fuzzy set of input quantity,representative heel data->Inputting membership value of the fuzzy set at p;
each cluster obtained after hierarchical clustering of the vehicle following data is further replaced in step 2.2.4 by calculating membership to obtain a weight multiplied by the membership, a similarity matrix is updated, and the similarity of the newly generated cluster and other clusters in the similarity matrix is calculated to further accelerate the hierarchical clustering speed;
step 4.2, aiming at the rules defined in the step 3, taking the cluster class of the vehicle following datamation partition as an input fuzzy set, wherein each input quantity corresponds to one rule, generating fuzzy rules, and if the same repeated rules of the previous parts exist in the rules, merging the extracted rules;
set the central value in each rule sequence number l asImproving the central value by using the confidence level of the input fuzzy set samples and the confidence level among the input fuzzy set samples, wherein the improved central value is +.>The calculation formula is as follows:
in doc (i) To input fuzzy set sample confidence, bel (l) The trust degree among input fuzzy set samples is that N is the number of the input fuzzy set samples by taking the cluster differentiated by the vehicle following data as the current rule sequence number index;
and (3) carrying out rule central value solving by using an improved central value formula, obtaining the confidence coefficient of an input sample and the confidence coefficient among samples by using vehicle relaxation data hierarchical clustering, replacing a central value formula obtained by using only simple processed original data corresponding to the weight of a fuzzy rule in a WM algorithm, and increasing the confidence coefficient for seeking a best solution.
Step 4.3, generating a fuzzy rule as follows:
IF x i1 isand…and x im is/>THEN y is B (l)
wherein B is (l) Is a Gaussian distribution membership function mu B(l) =μ(y·,y (l) ,σ (l) ) Is used to determine the fuzzy set of (a),gaussian distribution membership function mu B(l) =μ(y·,y (l) ,σ (l) ) Medium parameter sigma (l) The calculation is as follows:
wherein y is (l) For the rule center value of rule number l,the central value of the sequence number of the input fuzzy set with the rule sequence number of l, u (l) Obtaining m membership products of cluster class for hierarchical clustering of vehicle following data, and performing +.>Membership product of sequence number of input fuzzy set with rule sequence number of l +.>An index corresponding to the kth data of the output fuzzy aggregation data with the rule sequence number of l is represented; a is that ij Obtaining cluster class for hierarchical clustering of the vehicle following data, wherein m represents the number of output fuzzy sets of rule l;
calculating confidence doc of rule (l) The following are provided:
in the method, in the process of the invention,center value of sequence number of output fuzzy set for rule sequence number l, < >>The central value of the sequence number of the input fuzzy set with the rule sequence number of l is k, k is the kth output model, and t is the t input model;
calculating the confidence coefficient of the rule to evaluate the rationality degree of the rule, and indicating that no conflict exists between the merged rules;
step 4.4, obtaining a prediction model by using a central average deblurring method;
let the activation degree of rule number l be act (l), the formula of activation degree act (l) is as follows:
in the method, in the process of the invention,fuzzy set A with rule sequence number of l ij Membership degree of (3);
let each rule center point and the prediction point construct a prediction formula f (x), the prediction formula f (x) is as follows:
in the method, in the process of the invention,is a rule center with a rule sequence number of l, and M is the rule number;
by calculating the central value of each ruleDescribing rule center point +.>And calculating each new rule generated by the relation with the current vehicle following data x' by using a prediction formula f (x), finally obtaining a predicted following data result of the following vehicle at the next moment, adjusting the current running speed by a driver according to the predicted vehicle data at the next moment, immediately decelerating the vehicle if the vehicle is predicted to be in a rear-end collision, and keeping the vehicle following by the driver if the predicted data is always in a safe and stable state. By using the method to predict the vehicle following data subjected to the rule extraction processing, whether the following state is separated or the collision and rear-end collision problem occurs in the running process of the vehicle can be obtained in advance, and the obtained prediction result has higher reliability and safety.
The invention improves the WM method through hierarchical clustering, improves the performance of the vehicle following fuzzy rule base, and ensures the effectiveness of the output result of the following data. The vehicle following optimizing control method based on safety consideration is designed by establishing the vehicle following model, so that the problems of separation of the following state of the vehicle following model and collision rear-end collision are solved, and the safety and reliability of the vehicle following model are ensured.