Safety monitoring system and method based on big data vehicle-mounted intelligent driving system
By adopting big data on-board intelligent driving system on unmanned vehicles, collecting and predicting road intersections, traffic lights and obstacle data, and using BP neural network model to adjust vehicle driving speed, solving the problem of degradation in the performance of unmanned vehicle perception system and improving the Safety and response speed reduce the risk of accidents.
Patent Information
- Application Number
- CN202410600295.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-15
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-05-15
AI Technical Summary
When existing unmanned vehicles drive on the road, the performance of the perception system decreases over time, resulting in slower perception of road obstacles and conditions, which may lead to traffic accidents and cause losses to enterprises and traffic.
Using a safety monitoring method based on big data vehicle intelligent driving system, by setting up a module set of vehicle intelligent driving system, collecting road intersections, traffic lights and obstacle data, using BP neural network model to predict the perceived data, and determining whether the vehicle needs to be adjusted Driving speed to avoid accidents.
It improves the perceived accuracy and response speed of driverless vehicles to road conditions, reduces the possibility of accidents, and avoids traffic and corporate losses.
Smart Images

Figure CN118665504B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of abnormal situation alarm systems, and more specifically, particularly relates to a safety monitoring system and method based on a big data vehicle-mounted intelligent driving system. Background Art
[0002] The on-board intelligent driving system is a system that uses modern information technology to integrate multiple technologies such as computers, sensors, control, and communications to achieve intelligent control and management of vehicles, thereby improving the safety, efficiency, and comfort of vehicles. It usually includes multiple aspects such as intelligent assisted driving systems, automatic driving systems, and vehicle networking systems. These systems usually rely on various sensors, such as radar, cameras, lidar, GPS, etc., as well as powerful computing power and artificial intelligence algorithms to process large amounts of data and make fast and accurate decisions. With the continuous advancement and maturity of technology, on-board intelligent driving systems will be more widely used in the future, providing more convenience and security for driving. At the same time, this will also pose new challenges and requirements for traffic rules, laws and regulations, and traffic management.
[0003] Chinese patent CN112002102B discloses a safety monitoring method and system thereof, wherein the safety monitoring method comprises the following steps: determining a target driving route; driving according to the target driving route and acquiring travel data in real time; analyzing the travel data and generating instructions, which are instructions for continuing monitoring or alarming. The present application has the technical effect of safely monitoring abnormal events and abnormal driving events in a vehicle, and timely alarming abnormal situations, thereby avoiding casualties.
[0004] Existing unmanned vehicles often rely on the unmanned vehicle's onboard intelligent driving system to perceive and avoid road conditions or obstacles based on the perception results when driving on the road; however, the existing perception system often relies on various monitoring devices installed on the vehicle, which collect road images through the monitoring equipment and then analyze the road images. As time goes by, the performance of the driving system will gradually decline, and the perception of obstacles and conditions on the road will become slower. If the vehicle's driving speed cannot be adjusted in time, the unmanned vehicle is likely to cause a traffic accident, which will not only endanger traffic but also cause losses to the company. Summary of the invention
[0005] In response to the problems in the related technology, the present invention proposes a safety monitoring system and method based on a big data on-board intelligent driving system to overcome the above-mentioned technical problems existing in the existing related technology.
[0006] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:
[0007] The present invention is a safety monitoring method based on a big data vehicle-mounted intelligent driving system, comprising the following steps:
[0008] S1. Setting a vehicle-mounted intelligent driving system module set and a test area set, and collecting the number of road intersections and the number of traffic lights in each area, to obtain a road intersection number set and a traffic light number set; setting a regional road sub-section matrix according to the test area set and the road intersection number set; collecting obstacle quantity data of each road sub-section in the regional road sub-section matrix, to obtain a regional road sub-section obstacle data matrix;
[0009] S2, using the perception module in the vehicle intelligent driving system module set to continuously perceive the intersections, traffic lights and obstacles in each test area in the test area set, respectively, to obtain a first test result data set matrix, a second test result data set matrix and a third test result data set matrix; predicting the perception efficiency at future moments according to the first test result data set matrix, the second test result data set matrix and the third test result data set matrix, to obtain a final first BP neural network model, a final second BP neural network model and a final third BP neural network model;
[0010] S3, using the final first BP neural network model, the final second BP neural network model and the final third BP neural network model to respectively predict the intersection perception data, the traffic light perception data and the obstacle intersection perception data at the future moment, and obtain the first prediction data matrix, the second prediction data matrix and the third prediction data matrix; setting the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle, and determining whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle driving speed according to the first prediction data matrix, the second prediction data matrix, the third prediction data matrix, the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle;
[0011] By setting the module set of the on-board intelligent driving system, the types of modules that need to be considered are determined; since the role of each module and the time required for the work are different, the modules in the on-board intelligent driving system module set are analyzed separately in this solution; by considering the three aspects of intersections, traffic lights and obstacles, the situations that unmanned vehicles may encounter when driving on the road are comprehensively considered, making the monitoring more comprehensive and accurate, and reducing the possibility of accidents involving unmanned vehicles; by collecting the perception speed of intersections, traffic lights and obstacles by the perception modules in the on-board intelligent driving system module set, a perception data set is obtained, and the future perception data is predicted through the collected perception data set. According to the prediction results, the time when each module cooperates to finally control the vehicle, and the driving speed of the vehicle, it is determined whether the vehicle can smoothly avoid intersections, traffic lights and obstacles or take related measures according to the driving system, so that the vehicle will not be in danger or cause danger to other vehicles; once it is determined according to the prediction results that the vehicle will have an accident, the speed of the vehicle will be adjusted in time so that the vehicle can pass safely, thereby avoiding losses to the enterprise and the occurrence of traffic hazards.
[0012] Preferably, the S1 comprises the following steps:
[0013] S11, setting the vehicle intelligent driving system module set a = {a1, a2}, where a1 and a2 represent the perception module and the decision module respectively;
[0014] S12. Set the test area set represents the i-th test area set, b represents the total number of the set test areas; collect the test area set The number of road intersections and traffic lights in each area in the road intersection set b′={b1′,b2′,...,b i ′,...,b b ′} and the set of traffic lights c′={c1′,c′2,...,c i ′,...,c b ′},b i ′ represents the number of road intersections in the set i-th test area, c i ′ represents the number of traffic lights in the set i-th test area; according to the test area set And the number of road intersections b′={b1′,b2′,...,b i ′,...,b b '}Set regional road sub-section matrix as follows,
[0015]
[0016] in, represents the jth road subsection in the set i-th test area, Indicates the total number of road subsections in the set i-th test area;
[0017] Collect the road sub-section matrix of the area Obstacle quantity data of each road sub-section in the regional road sub-section is obtained, and the obstacle data matrix c of the regional road sub-section is obtained; as follows,
[0018]
[0019] Among them, c ij Indicates the number of obstacles in the j-th road sub-section in the set i-th test area;
[0020] By setting the test area set, the location for data collection is determined; by collecting data on the number of road intersections, the number of traffic lights, and the number of obstacles in each road sub-section in each area of the test area set, data support is provided for the subsequent perception of these road conditions through the perception module.
[0021] Preferably, S2 comprises the following steps:
[0022] S21, set the total test time to be f; divide the total test time f into time points to obtain a test time point set f i ′ represents the i-th test time point of the division, Indicates the total number of divided test time points;
[0023] S22, at the test time point At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set Continuously sense the intersections in each test area in the first test result data set matrix as follows,
[0024]
[0025] in, Indicates the test area set The result data set is obtained by performing continuous perception tests on each intersection in the i-th test area at the j-th test time point;
[0026] represents the distance between the vehicle and the kth road intersection when the kth road intersection in the ith test area is perceived at the jth test time point;
[0027] At the test time point set At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set The traffic lights in each test area are continuously sensed to obtain the second test result data set matrix g; as follows,
[0028]
[0029] Among them, g ij Indicates the test area set The result data set obtained by performing continuous perception tests on each traffic light in the i-th test area at the j-th test time point;
[0030] g ijk represents the distance between the vehicle and the kth traffic light when the vehicle perceives the kth traffic light in the ith test area at the jth test time point;
[0031] At the test time point set At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set The obstacles in each road sub-section in each test area are continuously sensed to obtain the third test result data set matrix g′; as follows,
[0032]
[0033] Among them, g i ' j Indicates the test area set The result data set is obtained by performing continuous perception tests on obstacles on each road sub-section in the i-th test area at the j-th test time point; g i ' jk It represents the minimum distance between the vehicle and the obstacle when the vehicle perceives the obstacle on the kth road sub-segment in the i-th test area at the j-th test time point;
[0034] S23, according to the first test result data set matrix The second test result data set matrix g and the third test result data set matrix g′ predict the perceived efficiency at future moments;
[0035] By dividing the time points, at each time point the perception module continuously perceives the intersections in each test area set, the traffic lights in each test area set, and the obstacles in each road sub-section in each test area to obtain perception data, thereby providing data support for the subsequent prediction of perception data at each future time point.
[0036] Preferably, the S23 comprises the following steps:
[0037] S231, setting the first training ratio value The second training ratio is The third training ratio is The first test ratio is The second test ratio is And the third test ratio value is According to the training ratio value Second training ratio value The third training ratio The first test ratio is The second test ratio is And the third test ratio value is The first test result data set matrix The second test result data set matrix g and the third test result data set matrix g′ are divided into data to obtain the first test result training data set matrix, the first test result test data set matrix, the second test result training data set matrix, the second test result test data set matrix, the third test result training data set matrix, the third test result test data set matrix, the first to-be-predicted result data set matrix, the second to-be-predicted result data set matrix and the third to-be-predicted result data set matrix;
[0038] S232, constructing a first BP neural network model, a second BP neural network model and a third BP neural network model; setting the number of input nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model to be and The number of hidden nodes are h1, h2 and h3 and the number of output nodes are h1′, h2′ and h3′ respectively;
[0039] The functions adopted by the input nodes and output nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model are all Purline functions, and the functions adopted by the hidden nodes are all Sigmoid functions;
[0040] The initial weights of the first BP neural network model are set to and the initial threshold is are respectively recorded as the first initial weight and the first initial threshold; the initial weight of the second BP neural network model is set to and the initial threshold is They are respectively recorded as the second initial weight and the second initial threshold; the initial weight of the third BP neural network model is set to l1 and the initial threshold is set to l2, which are respectively recorded as the third initial weight and the third initial threshold;
[0041] S233, inputting the first test result training data set matrix, the second test result training data set matrix and the third test result training data set matrix into the input nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model respectively for training operations, to obtain the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model;
[0042] S234, inputting the first test result test data set matrix, the second test result test data set matrix and the third test result test data set matrix into the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model for testing respectively, and after the testing is completed, obtaining the final first BP neural network model, the final second BP neural network model and the final third BP neural network model;
[0043] A BP neural network model is used to predict the collected perception data; first, the perception data is divided into a training set, a test set, and a set to be predicted, the BP neural network model is trained with the training set, and then the trained BP neural network model is tested with the test set, thereby obtaining a BP neural network model that can accurately predict the perception data, thereby improving the accuracy of safety monitoring of the driving system.
[0044] Preferably, the S233 comprises the following steps:
[0045] S2331, setting a first training error threshold, a second training error threshold, and a third training error threshold;
[0046] S2332, inputting the first test result training data set matrix into the input node of the first BP neural network model for training, and when the training error of the first BP neural network model is less than the first training error threshold, stopping the training to obtain the trained first BP neural network model;
[0047] Inputting the second test result training data set matrix into the input node of the second BP neural network model for training, and when the training error of the second BP neural network model is less than the second training error threshold, stopping the training to obtain the trained second BP neural network model;
[0048] Inputting the third test result training data set matrix into the input node of the third BP neural network model for training, and when the training error of the third BP neural network model is less than the third training error threshold, stopping the training to obtain the trained third BP neural network model;
[0049] The BP neural network model is trained through the training set, so that the prediction efficiency and accuracy of the BP neural network model for the perception data are improved.
[0050] Preferably, the S234 comprises the following steps:
[0051] S2341, inputting the first test result test data set matrix, the second test result test data set matrix and the third test result test data set matrix into the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model for testing, and obtaining the first test accuracy m1, the second test accuracy m2 and the third test accuracy m3 respectively after the test is completed;
[0052] S2342, setting a first accuracy threshold m1′, a second accuracy threshold m′2, and a third accuracy threshold m3′;
[0053] When m1≥m1′, the trained first BP neural network model is used as the final first BP neural network model; otherwise, the first initial weight of the trained first BP neural network model is and the first initial threshold Perform adjustment optimization to obtain the optimal first weight and the optimal first threshold
[0054] When m2≥m′2, the trained second BP neural network model is used as the final second BP neural network model; otherwise, the second initial weight of the trained second BP neural network model is and the second initial threshold Perform adjustment optimization to obtain the optimal second weight and the optimal second threshold
[0055] When m3≥m3′, the trained third BP neural network model is used as the final third BP neural network model; otherwise, the third initial weight l1 and the third initial threshold l2 of the trained third BP neural network model are adjusted and optimized to obtain the optimal third weight n1 and the optimal third threshold n2;
[0056] The trained BP neural network model is tested using the test set perception data. When the test accuracy does not meet the requirements, the relevant parameters of the BP neural network model are adjusted to further improve the accuracy of the BP neural network model in predicting the perception data.
[0057] Preferably, the first initial weight of the trained first BP neural network model and the first initial threshold Performing adjustment optimization and adjusting the second initial weight of the trained second BP neural network model and the second initial threshold The adjustment and optimization and the adjustment and optimization of the third initial weight l1 and the third initial threshold l2 of the trained third BP neural network model specifically include the following steps:
[0058] S23421, construct the first day cattle population, the second day cattle population and the third day cattle population; set the scales of the first day cattle population, the second day cattle population and the third day cattle population to be l1′, l2′ and l3′ respectively, then the first day cattle population, the second day cattle population and the third day cattle population are respectively expressed as and and They represent the i-th beetle in the first beetle population, the i-th beetle in the second beetle population, and the i-th beetle in the third beetle population, respectively;
[0059] Set the first maximum number of iterations to The second maximum number of iterations is And the third maximum number of iterations is The optimization search space dimensions of the first day beetle population, the second day beetle population and the third day beetle population are all set to be two-dimensional;
[0060] S23422: According to the first initial weight and the first initial threshold Randomly generate the initial position set of the first day cattle population According to the second initial weight and the second initial threshold Randomly generate the initial position set of the cattle population on the second day The third initial position set of the Tianniu population is randomly generated according to the third initial weight l1 and the third initial threshold l2 n i ′1 and n i ′2 respectively represents the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the first longhorn beetle population; and Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the second longhorn beetle population; and Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the third longhorn beetle population;
[0061] The fitness functions of the first beetle population, the second beetle population, and the third beetle population are set as the first fitness function, the second fitness function, and the third fitness function, respectively; the function formulas of the first fitness function, the second fitness function, and the third fitness function are respectively as follows:
[0062]
[0063] Wherein, o1, o2 and o3 represent the first fitness function, the second fitness function and the third fitness function respectively; o1′, o′2 and o3′ represent the first bias, the second bias and the third bias respectively;
[0064] S23423, start iterative operation, set the first current iteration number Second current iteration number and the third current iteration number In each round of iteration, the positions of the individual beetles in the first beetle population initial position set, the second beetle population initial position set, and the third beetle population initial position set are updated according to the function formulas of the first fitness function, the second fitness function, and the third fitness function; and in each round of iteration, the global best beetle individual and the global best fitness in the first beetle population, the second beetle population, and the third beetle population are obtained;
[0065] S23424, when When the first beetle population stops the iteration operation, the optimal first weight value is output and the optimal first threshold when When , the second cattle population stops the iteration operation and outputs the optimal second weight and the optimal second threshold when When , the third beetle population stops the iterative operation and outputs the optimal third weight n1 and the optimal third threshold n2;
[0066] S23425, the optimal first weight and the optimal first threshold Substitute the trained first BP neural network model to obtain the final first BP neural network model; replace the optimal second weight and the optimal second threshold Substituting the trained second BP neural network model into the final second BP neural network model; substituting the optimal third weight n1 and the optimal third threshold n2 into the trained third BP neural network model to obtain the final third BP neural network model;
[0067] The beetle swarm algorithm is used to tune the parameters of the BP neural network model, so that multiple beetles in the beetle population are adjusted simultaneously, which improves the tuning efficiency; and the initial position of the beetle is generated randomly to avoid the local optimal situation.
[0068] Preferably, S3 comprises the following steps:
[0069] S31, using the final first BP neural network model to predict the data of the first result data set matrix to be predicted at a future time, to obtain a first prediction data matrix p1; using the final second BP neural network model to predict the data of the second result data set matrix to be predicted at a future time, to obtain a second prediction data matrix p2; using the final third BP neural network model to predict the data of the third result data set matrix to be predicted at a future time, to obtain a third prediction data matrix p3;
[0070] The first prediction data matrix p1, the second prediction data matrix p2 and the third prediction data matrix p3 are respectively as follows:
[0071]
[0072]
[0073] Among them, p 1ij It indicates the perception data set of each intersection in the i-th test area at the j-th time point in the future. p 1ijk represents the predicted distance between the vehicle and the kth intersection when the vehicle perceives the kth intersection in the ith test area at the jth time point in the future; p 2ij It indicates the perception data set of each traffic light in the i-th test area at the j-th time point in the future. p 2ijk represents the predicted distance between the vehicle and the kth traffic light when the vehicle perceives the kth traffic light in the ith test area at the jth time point in the future; p3ij It represents the perception dataset that predicts the obstacles of each road sub-section in the i-th test area at the j-th time point in the future. p 3ijk represents the predicted minimum distance between the vehicle and the obstacle when the vehicle perceives the obstacle in the i-th test area at the j-th time point in the future; p′ represents the total number of time points set for prediction in the future;
[0074] S32, setting the sum of the time required for the perception module to transmit the intersection perception data, the traffic light perception data and the obstacle perception data to the decision module, the time required for the decision module to make a decision and the time required for the decision module to adjust the vehicle to be and They are recorded as the first decision-making time, the second decision-making time and the third decision-making time respectively; the current speed of the vehicle is set to
[0075] S33, according to the first prediction data matrix p1, the second prediction data matrix p2, the third prediction data matrix p3, the first decision consumption time Second decision-making time Third decision-making time and the vehicle's current speed Determine whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle speed;
[0076] By combining these factors based on the time required for perception, the vehicle's driving speed, and the time required for the decision-making module, it is possible to accurately determine whether the vehicle needs to be adjusted, thereby improving the comprehensiveness and accuracy of the judgment.
[0077] Preferably, the S33 comprises the following steps:
[0078] S331, when or or When the vehicle driving system needs to be warned, otherwise, the vehicle driving system does not need to be warned.
[0079] S332: After the vehicle driving system issues an alarm, the vehicle driving system needs to reduce the vehicle's driving speed so that and and
[0080] The above conditions can be used to quantify the process of determining whether the vehicle needs adjustment, making the determination process more convenient and more reliable.
[0081] A safety monitoring system based on a big data vehicle-mounted intelligent driving system, comprising a road intersection data acquisition module, a traffic light number data acquisition module, an obstacle data acquisition module, a first prediction module, a second prediction module and a final determination module;
[0082] The road intersection data collection module is used to collect the number of road intersections in each area and obtain a first test result data set matrix;
[0083] The traffic light number data collection module is used to collect the number of traffic lights in each area and obtain a second test result data set matrix;
[0084] The obstacle data collection module is used to collect obstacle quantity data of each road sub-section in each area and obtain a third test result data set matrix;
[0085] The first prediction module is used to predict the perceived efficiency at a future moment according to the first test result data set matrix, the second test result data set matrix, and the third test result data set matrix;
[0086] The second prediction module is used to use the final first BP neural network model, the final second BP neural network model and the final third BP neural network model to respectively predict the intersection perception data, the traffic light perception data and the obstacle intersection perception data at the future moment;
[0087] The final judgment module is used to determine whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle driving speed based on the first prediction data matrix, the second prediction data matrix, the third prediction data matrix, the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle.
[0088] The present invention has the following beneficial effects:
[0089] 1. In the present invention, a road intersection data acquisition module, a traffic light number data acquisition module, an obstacle data acquisition module, a first prediction module, a second prediction module and a final judgment module are set; by setting a vehicle-mounted intelligent driving system module set, the types of modules that need to be considered are determined; since the function of each module and the time required for the work are different, the modules in the vehicle-mounted intelligent driving system module set are analyzed separately in this solution; by considering the three aspects of intersections, traffic lights and obstacles, the situations that unmanned vehicles may encounter when driving on the road are fully considered, so that the monitoring is more comprehensive and accurate, and the possibility of accidents involving unmanned vehicles is reduced; by The perception speed of intersections, traffic lights and obstacles of the perception module concentrated in the on-board intelligent driving system module is collected to obtain a perception data set. The future perception data is predicted through the collected perception data set. According to the prediction results, the time for each module to cooperate to finally control the vehicle and the vehicle's driving speed, it is determined whether the vehicle can successfully avoid intersections, traffic lights and obstacles or take related measures according to the driving system, so that the vehicle will not be in danger or cause danger to other vehicles. Once it is determined according to the prediction results that the vehicle will have an accident, the speed of the vehicle will be adjusted in time to allow the vehicle to pass safely, thereby avoiding losses to the company and harm to traffic.
[0090] 2. In the present invention, the location for collecting data is determined by setting a test area set; by collecting data on the number of road intersections, the number of traffic lights, and the number of obstacles in each road sub-section in each area of the test area set, data support is provided for the subsequent perception of these road conditions by the perception module; at the same time, by dividing the time points, at each time point, the perception module continuously perceives the intersections in each test area in the test area set, the traffic lights in each test area in the test area set, and the obstacles in each road sub-section in each test area to obtain perception data, thereby providing data support for the subsequent prediction of the perception data at each future time point.
[0091] 3. In the present invention, the collected perception data is predicted by adopting a BP neural network model; first, the perception data is divided into a training set, a test set and a set to be predicted, the BP neural network model is trained by the training set, and the trained BP neural network model is tested by the test set, so as to obtain a BP neural network model that can accurately predict the perception data, thereby improving the accuracy of safety monitoring of the driving system, wherein the parameters of the BP neural network model are tuned by a longhorn beetle swarm algorithm, so that multiple longhorn beetles in the longhorn beetle population are adjusted simultaneously, thereby improving the tuning efficiency; and the position of the initial longhorn beetle is randomly generated, thereby avoiding the occurrence of a local optimal situation.
[0092] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS
[0093] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.
[0094] Figure 1 The present invention is a flowchart of a safety monitoring system for a big data-based on-vehicle intelligent driving system monitoring an on-vehicle intelligent driving system. DETAILED DESCRIPTION
[0095] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.
[0096] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.
[0097] The present invention is a safety monitoring method based on a big data vehicle-mounted intelligent driving system, comprising the following steps:
[0098] S1. Setting a vehicle-mounted intelligent driving system module set and a test area set, and collecting the number of road intersections and the number of traffic lights in each area, to obtain a road intersection number set and a traffic light number set; setting a regional road sub-section matrix according to the test area set and the road intersection number set; collecting obstacle quantity data of each road sub-section in the regional road sub-section matrix, to obtain a regional road sub-section obstacle data matrix;
[0099] The S1 comprises the following steps:
[0100] S11, setting the vehicle intelligent driving system module set a = {a1, a2}, where a1 and a2 represent the perception module and the decision module respectively;
[0101] S12. Set the test area set represents the i-th test area set, b represents the total number of the test areas set; collect the test area set The number of road intersections and traffic lights in each area in the road intersection set b′={b1′,b2′,...,b i ′,...,b b ′} and the set of traffic lights c′={c1′,c′2,...,c i ′,...,c b ′},b i ′ represents the number of road intersections in the set i-th test area, c i ′ represents the number of traffic lights in the set i-th test area; according to the test area set And the number of road intersections b′={b1′,b2′,...,b i ′,...,b b '}Set regional road sub-section matrix as follows,
[0102]
[0103] in, represents the jth road subsection in the set i-th test area, Indicates the total number of road subsections in the set i-th test area;
[0104] Collect the road sub-section matrix of the area Obstacle quantity data of each road sub-section in the regional road sub-section is obtained, and the obstacle data matrix c of the regional road sub-section is obtained; as follows,
[0105]
[0106] Among them, c ij Indicates the number of obstacles in the j-th road sub-section in the set i-th test area;
[0107] S2, using the perception module in the vehicle intelligent driving system module set to continuously perceive the intersections, traffic lights and obstacles in each test area in the test area set, respectively, to obtain a first test result data set matrix, a second test result data set matrix and a third test result data set matrix; predicting the perception efficiency at future moments according to the first test result data set matrix, the second test result data set matrix and the third test result data set matrix, to obtain a final first BP neural network model, a final second BP neural network model and a final third BP neural network model;
[0108] The S2 comprises the following steps:
[0109] S21, set the total test time to be f; divide the total test time f into time points to obtain a test time point set f i ′ represents the i-th test time point of the division, Indicates the total number of divided test time points;
[0110] S22, at the test time point At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set Continuously sense the intersections in each test area in the first test result data set matrix as follows,
[0111]
[0112] in, Indicates the test area set The result data set is obtained by performing continuous perception tests on each intersection in the i-th test area at the j-th test time point;
[0113] represents the distance between the vehicle and the kth road intersection when the kth road intersection in the ith test area is perceived at the jth test time point;
[0114] At the test time point set At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set The traffic lights in each test area are continuously sensed to obtain the second test result data set matrix g; as follows,
[0115]
[0116] Among them, g ij Indicates the test area set The result data set obtained by performing continuous perception tests on each traffic light in the i-th test area at the j-th test time point;
[0117] g ijk represents the distance between the vehicle and the kth traffic light when the vehicle perceives the kth traffic light in the ith test area at the jth test time point;
[0118] At the test time point set At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set The obstacles in each road sub-section in each test area are continuously sensed to obtain the third test result data set matrix g′; as follows,
[0119]
[0120] Among them, g i ' j Indicates the test area set The result data set is obtained by performing continuous perception tests on obstacles on each road sub-section in the i-th test area at the j-th test time point; g i ' jk It represents the minimum distance between the vehicle and the obstacle when the vehicle perceives the obstacle on the kth road sub-segment in the i-th test area at the j-th test time point;
[0121] S23, according to the first test result data set matrix The second test result data set matrix g and the third test result data set matrix g′ predict the perceived efficiency at future moments;
[0122] The S23 comprises the following steps:
[0123] S231, setting the first training ratio value The second training ratio is The third training ratio is The first test ratio is The second test ratio is And the third test ratio value is According to the training ratio value Second training ratio value The third training ratio The first test ratio is The second test ratio is And the third test ratio value is For the first test result data set matrix The second test result data set matrix g and the third test result data set matrix g′ are divided into data to obtain the first test result training data set matrix, the first test result test data set matrix, the second test result training data set matrix, the second test result test data set matrix, the third test result training data set matrix, the third test result test data set matrix, the first to-be-predicted result data set matrix, the second to-be-predicted result data set matrix and the third to-be-predicted result data set matrix;
[0124] S232, constructing a first BP neural network model, a second BP neural network model and a third BP neural network model; setting the number of input nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model to be and The number of hidden nodes are h1, h2 and h3 and the number of output nodes are h1′, h2′ and h3′ respectively;
[0125] The functions adopted by the input nodes and output nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model are all Purline functions, and the functions adopted by the hidden nodes are all Sigmoid functions;
[0126] The initial weights of the first BP neural network model are set to and the initial threshold is are respectively recorded as the first initial weight and the first initial threshold; the initial weight of the second BP neural network model is set to and the initial threshold is They are respectively recorded as the second initial weight and the second initial threshold; the initial weight of the third BP neural network model is set to l1 and the initial threshold is set to l2, which are respectively recorded as the third initial weight and the third initial threshold;
[0127] S233, inputting the first test result training data set matrix, the second test result training data set matrix and the third test result training data set matrix into the input nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model respectively for training operations, to obtain the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model;
[0128] The S233 comprises the following steps:
[0129] S2331, setting a first training error threshold, a second training error threshold, and a third training error threshold;
[0130] S2332, inputting the first test result training data set matrix into the input node of the first BP neural network model for training, and when the training error of the first BP neural network model is less than the first training error threshold, stopping the training to obtain the trained first BP neural network model;
[0131] Inputting the second test result training data set matrix into the input node of the second BP neural network model for training, and when the training error of the second BP neural network model is less than the second training error threshold, stopping the training to obtain the trained second BP neural network model;
[0132] Inputting the third test result training data set matrix into the input node of the third BP neural network model for training, and when the training error of the third BP neural network model is less than the third training error threshold, stopping the training to obtain the trained third BP neural network model;
[0133] S234, inputting the first test result test data set matrix, the second test result test data set matrix and the third test result test data set matrix into the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model for testing respectively, and after the testing is completed, obtaining the final first BP neural network model, the final second BP neural network model and the final third BP neural network model;
[0134] The S234 comprises the following steps:
[0135] S2341, inputting the first test result test data set matrix, the second test result test data set matrix and the third test result test data set matrix into the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model for testing, and obtaining the first test accuracy m1, the second test accuracy m2 and the third test accuracy m3 respectively after the test is completed;
[0136] S2342, setting a first accuracy threshold m1′, a second accuracy threshold m′2, and a third accuracy threshold m3′;
[0137] When m1≥m1′, the trained first BP neural network model is used as the final first BP neural network model; otherwise, the first initial weight of the trained first BP neural network model is and the first initial threshold Perform adjustment optimization to obtain the optimal first weight and the optimal first threshold
[0138] When m2≥m′2, the trained second BP neural network model is used as the final second BP neural network model; otherwise, the second initial weight of the trained second BP neural network model is and the second initial threshold Perform adjustment optimization to obtain the optimal second weight and the optimal second threshold
[0139] When m3≥m3′, the trained third BP neural network model is used as the final third BP neural network model; otherwise, the third initial weight l1 and the third initial threshold l2 of the trained third BP neural network model are adjusted and optimized to obtain the optimal third weight n1 and the optimal third threshold n2;
[0140] The first initial weight of the trained first BP neural network model and the first initial threshold Performing adjustment optimization and adjusting the second initial weight of the trained second BP neural network model and the second initial threshold The adjustment and optimization and the adjustment and optimization of the third initial weight l1 and the third initial threshold l2 of the trained third BP neural network model specifically include the following steps:
[0141] S23421, construct the first day cattle population, the second day cattle population and the third day cattle population; set the scales of the first day cattle population, the second day cattle population and the third day cattle population to be l1′, l2′ and l3′ respectively, then the first day cattle population, the second day cattle population and the third day cattle population are respectively expressed as and and They represent the i-th beetle in the first beetle population, the i-th beetle in the second beetle population, and the i-th beetle in the third beetle population, respectively;
[0142] Set the first maximum number of iterations to The second maximum number of iterations is And the third maximum number of iterations is The optimization search space dimensions of the first day beetle population, the second day beetle population and the third day beetle population are all set to be two-dimensional;
[0143] S23422: According to the first initial weight and the first initial threshold Randomly generate the initial position set of the first day cattle population According to the second initial weight and the second initial threshold Randomly generate the initial position set of the cattle population on the second day The third initial position set of the Tianniu population is randomly generated according to the third initial weight l1 and the third initial threshold l2 n i ′1 and n i ′2 respectively represents the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the first longhorn beetle population; and Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the second longhorn beetle population; and Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the third longhorn beetle population;
[0144] The fitness functions of the first beetle population, the second beetle population, and the third beetle population are set as the first fitness function, the second fitness function, and the third fitness function, respectively; the function formulas of the first fitness function, the second fitness function, and the third fitness function are respectively as follows:
[0145]
[0146] Wherein, o1, o2 and o3 represent the first fitness function, the second fitness function and the third fitness function respectively; o1′, o′2 and o3′ represent the first bias, the second bias and the third bias respectively;
[0147] S23423, start iterative operation, set the first current iteration number Second current iteration number and the third current iteration number In each round of iteration, the positions of the individual beetles in the first beetle population initial position set, the second beetle population initial position set, and the third beetle population initial position set are updated according to the function formulas of the first fitness function, the second fitness function, and the third fitness function; and in each round of iteration, the global best beetle individual and the global best fitness in the first beetle population, the second beetle population, and the third beetle population are obtained;
[0148] S23424, when When the first beetle population stops the iteration operation, the optimal first weight value is output and the optimal first threshold when When , the second cattle population stops the iteration operation and outputs the optimal second weight and the optimal second threshold when When , the third beetle population stops the iterative operation and outputs the optimal third weight n1 and the optimal third threshold n2;
[0149] S23425, the optimal first weight and the optimal first threshold Substitute the trained first BP neural network model to obtain the final first BP neural network model; replace the optimal second weight and the optimal second threshold Substituting the trained second BP neural network model into the final second BP neural network model; substituting the optimal third weight n1 and the optimal third threshold n2 into the trained third BP neural network model to obtain the final third BP neural network model;
[0150] S3, using the final first BP neural network model, the final second BP neural network model and the final third BP neural network model to respectively predict the intersection perception data, the traffic light perception data and the obstacle intersection perception data at the future moment, and obtain the first prediction data matrix, the second prediction data matrix and the third prediction data matrix; setting the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle, and determining whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle driving speed according to the first prediction data matrix, the second prediction data matrix, the third prediction data matrix, the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle;
[0151] The S3 comprises the following steps:
[0152] S31, using the final first BP neural network model to predict the data of the first result data set matrix to be predicted at a future time, to obtain a first prediction data matrix p1; using the final second BP neural network model to predict the data of the second result data set matrix to be predicted at a future time, to obtain a second prediction data matrix p2; using the final third BP neural network model to predict the data of the third result data set matrix to be predicted at a future time, to obtain a third prediction data matrix p3;
[0153] The first prediction data matrix p1, the second prediction data matrix p2 and the third prediction data matrix p3 are respectively as follows:
[0154]
[0155]
[0156] Among them, p 1ij It indicates the perception data set of each intersection in the i-th test area at the j-th time point in the future. p 1ijk represents the predicted distance between the vehicle and the kth intersection when the vehicle perceives the kth intersection in the ith test area at the jth time point in the future; p 2ij It indicates the perception data set of each traffic light in the i-th test area at the j-th time point in the future. p 2ijkrepresents the predicted distance between the vehicle and the kth traffic light when the vehicle perceives the kth traffic light in the ith test area at the jth time point in the future; p 3ij It represents the perception dataset that predicts the obstacles of each road sub-section in the i-th test area at the j-th time point in the future. p 3ijk represents the predicted minimum distance between the vehicle and the obstacle when the vehicle perceives the obstacle in the i-th test area at the j-th time point in the future; p′ represents the total number of time points set for prediction in the future;
[0157] S32, setting the sum of the time required for the perception module to transmit the intersection perception data, the traffic light perception data and the obstacle perception data to the decision module, the time required for the decision module to make a decision and the time required for the decision module to adjust the vehicle to be and They are recorded as the first decision-making time, the second decision-making time and the third decision-making time respectively; the current speed of the vehicle is set to
[0158] S33, according to the first prediction data matrix p1, the second prediction data matrix p2, the third prediction data matrix p3, the first decision consumption time Second decision-making time Third decision-making time and the vehicle's current speed Determine whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle speed;
[0159] The S33 comprises the following steps:
[0160] S331, when or or When the vehicle driving system needs to be warned, otherwise, the vehicle driving system does not need to be warned.
[0161] S332: After the vehicle driving system issues an alarm, the vehicle driving system needs to reduce the vehicle's driving speed so that and and
[0162] A safety monitoring system based on a big data vehicle-mounted intelligent driving system, comprising a road intersection data acquisition module, a traffic light number data acquisition module, an obstacle data acquisition module, a first prediction module, a second prediction module and a final determination module;
[0163] The road intersection data collection module is used to collect the number of road intersections in each area and obtain a first test result data set matrix;
[0164] The traffic light number data collection module is used to collect the number of traffic lights in each area and obtain a second test result data set matrix;
[0165] The obstacle data collection module is used to collect obstacle quantity data of each road sub-section in each area and obtain a third test result data set matrix;
[0166] The first prediction module is used to predict the perceived efficiency at a future moment according to the first test result data set matrix, the second test result data set matrix, and the third test result data set matrix;
[0167] The second prediction module is used to use the final first BP neural network model, the final second BP neural network model and the final third BP neural network model to respectively predict the intersection perception data, the traffic light perception data and the obstacle intersection perception data at the future moment;
[0168] The final judgment module is used to determine whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle driving speed based on the first prediction data matrix, the second prediction data matrix, the third prediction data matrix, the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle.
[0169] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0170] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.
Claims
1. A safety monitoring method based on a big data vehicle-mounted intelligent driving system, characterized in that: The following steps are involved: S1. Setting a vehicle-mounted intelligent driving system module set and a test area set, and collecting the number of road intersections and the number of traffic lights in each area, to obtain a road intersection number set and a traffic light number set; setting a regional road sub-section matrix according to the test area set and the road intersection number set; collecting obstacle quantity data of each road sub-section in the regional road sub-section matrix, to obtain a regional road sub-section obstacle data matrix; S2, using the perception module in the vehicle intelligent driving system module set to continuously perceive the intersections, traffic lights and obstacles in each test area in the test area set, to obtain a first test result data set matrix, a second test result data set matrix and a third test result data set matrix; Predicting the perceived efficiency at future moments according to the first test result data set matrix, the second test result data set matrix, and the third test result data set matrix to obtain a final first BP neural network model, a final second BP neural network model, and a final third BP neural network model; S3, using the final first BP neural network model, the final second BP neural network model and the final third BP neural network model to respectively predict the intersection perception data, the traffic light perception data and the obstacle intersection perception data at the future moment, to obtain a first prediction data matrix, a second prediction data matrix and a third prediction data matrix; The first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle are set, and whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle driving speed according to the first prediction data matrix, the second prediction data matrix, the third prediction data matrix, the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle.
2. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 1 is characterized in that: The S1 comprises the following steps: S11, setting the vehicle intelligent driving system module set a = {a1, a2}, where a1 and a2 represent the perception module and the decision module respectively; S12. Set the test area set represents the i-th test area set, b represents the total number of the test areas set; collect the test area set The number of road intersections and traffic lights in each area in the road intersection set b′={b1′,b2′,...,b i ′,...,b b ′} and the set of traffic lights c′={c1′,c′2,...,c i ′,...,c b ′},b i ′ represents the number of road intersections in the set i-th test area, c i ′ represents the number of traffic lights in the set i-th test area; according to the test area set And the number of road intersections b′={b1′,b2′,...,b i ′,...,b b '}Set regional road sub-section matrix as follows, in, represents the jth road subsection in the set i-th test area, Indicates the total number of road subsections in the set i-th test area; Collect the road sub-section matrix of the area Obstacle quantity data of each road sub-section in the regional road sub-section is obtained, and the obstacle data matrix c of the regional road sub-section is obtained; as follows, Among them, c ij Indicates the number of obstacles in the j-th road sub-section in the set i-th test area.
3. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 1 is characterized in that: The S2 comprises the following steps: S21, set the total test time to be f; divide the total test time f into time points to obtain a test time point set f i ′ represents the i-th test time point of the division, Indicates the total number of divided test time points; S22, at the test time point At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set Continuously sense the intersections in each test area in the first test result data set matrix as follows, in, Indicates the test area set The result data set is obtained by performing continuous perception tests on each intersection in the i-th test area at the j-th test time point; represents the distance between the vehicle and the kth road intersection when the kth road intersection in the ith test area is perceived at the jth test time point; At the test time point set At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set The traffic lights in each test area are continuously sensed to obtain the second test result data set matrix g; as follows, Among them, g ij Indicates the test area set The result data set obtained by performing continuous perception tests on each traffic light in the i-th test area at the j-th test time point; g ij = {g ij1 ,g ij2 ,...,g ijk ,...,g ijci′ }, g ijk represents the distance between the vehicle and the kth traffic light when the vehicle perceives the kth traffic light in the ith test area at the jth test time point; At the test time point set At each test time point, the perception modules in the vehicle intelligent driving system module set a = {a1, a2, a3} are used to detect the test area set The obstacles in each road sub-section in each test area are continuously sensed to obtain the third test result data set matrix g′; as follows, Among them, g i ' j Indicates the test area set The result dataset is obtained by performing continuous perception tests on obstacles on each road sub-section in the i-th test area at the j-th test time point; g i ' j = {g i ' j1 ,g i ' j2 ,...,g i ' jk ,...,g i ' jb~i }, g i ' jk It represents the minimum distance between the vehicle and the obstacle when the vehicle perceives the obstacle on the kth road sub-segment in the i-th test area at the j-th test time point; S23, according to the first test result data set matrix The second test result data set matrix g and the third test result data set matrix g′ predict the perceived efficiency at future moments.
4. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 3 is characterized in that: The S23 comprises the following steps: S231, setting the first training ratio value The second training ratio is The third training ratio is The first test ratio is The second test ratio is And the third test ratio value is According to the training ratio value Second training ratio value The third training ratio The first test ratio is The second test ratio is And the third test ratio value is For the first test result data set matrix The second test result data set matrix g and the third test result data set matrix g′ are divided into data to obtain the first test result training data set matrix, the first test result test data set matrix, the second test result training data set matrix, the second test result test data set matrix, the third test result training data set matrix, the third test result test data set matrix, the first to-be-predicted result data set matrix, the second to-be-predicted result data set matrix and the third to-be-predicted result data set matrix; S232, constructing a first BP neural network model, a second BP neural network model and a third BP neural network model; setting the number of input nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model to be and The number of hidden nodes are h1, h2 and h3 and the number of output nodes are h1′, h2′ and h3′ respectively; The functions adopted by the input nodes and output nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model are all Purline functions, and the functions adopted by the hidden nodes are all Sigmoid functions; The initial weights of the first BP neural network model are set to and the initial threshold is are respectively recorded as the first initial weight and the first initial threshold; the initial weight of the second BP neural network model is set to and the initial threshold is They are respectively recorded as the second initial weight and the second initial threshold; the initial weight of the third BP neural network model is set to l1 and the initial threshold is set to l2, which are respectively recorded as the third initial weight and the third initial threshold; S233, inputting the first test result training data set matrix, the second test result training data set matrix and the third test result training data set matrix into the input nodes of the first BP neural network model, the second BP neural network model and the third BP neural network model respectively for training operations, to obtain the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model; S234. Input the first test result test data set matrix, the second test result test data set matrix and the third test result test data set matrix into the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model respectively for testing. After the test is completed, the final first BP neural network model, the final second BP neural network model and the final third BP neural network model are obtained.
5. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 4 is characterized in that: The S233 comprises the following steps: S2331, setting a first training error threshold, a second training error threshold, and a third training error threshold; S2332, inputting the first test result training data set matrix into the input node of the first BP neural network model for training, and when the training error of the first BP neural network model is less than the first training error threshold, stopping the training to obtain the trained first BP neural network model; Inputting the second test result training data set matrix into the input node of the second BP neural network model for training, and when the training error of the second BP neural network model is less than the second training error threshold, stopping the training to obtain the trained second BP neural network model; The third test result training data set matrix is input into the input node of the third BP neural network model for training. When the training error of the third BP neural network model is less than the third training error threshold, the training is stopped to obtain the trained third BP neural network model.
6. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 4 is characterized in that: The S234 comprises the following steps: S2341, inputting the first test result test data set matrix, the second test result test data set matrix and the third test result test data set matrix into the trained first BP neural network model, the trained second BP neural network model and the trained third BP neural network model for testing, and obtaining the first test accuracy m1, the second test accuracy m2 and the third test accuracy m3 respectively after the test is completed; S2342, setting a first accuracy threshold m1′, a second accuracy threshold m′2, and a third accuracy threshold m3′; When m1≥m1′, the trained first BP neural network model is used as the final first BP neural network model; otherwise, the first initial weight of the trained first BP neural network model is and the first initial threshold Perform adjustment optimization to obtain the optimal first weight and the optimal first threshold When m2≥m′2, the trained second BP neural network model is used as the final second BP neural network model; otherwise, the second initial weight h~1 and the second initial threshold h~2 of the trained second BP neural network model are adjusted and optimized to obtain the optimal second weight and the optimal second threshold When m3≥m3′, the trained third BP neural network model is used as the final third BP neural network model; otherwise, the third initial weight l1 and the third initial threshold l2 of the trained third BP neural network model are adjusted and optimized to obtain the optimal third weight n1 and the optimal third threshold n2. 2。 7. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 6 is characterized in that: The first initial weight of the trained first BP neural network model and the first initial threshold Performing adjustment optimization and adjusting the second initial weight of the trained second BP neural network model and the second initial threshold The adjustment and optimization and the adjustment and optimization of the third initial weight l1 and the third initial threshold l2 of the trained third BP neural network model specifically include the following steps: S23421, construct the first day cattle population, the second day cattle population and the third day cattle population; set the scales of the first day cattle population, the second day cattle population and the third day cattle population to be l1′, l2′ and l3′ respectively, then the first day cattle population, the second day cattle population and the third day cattle population are respectively expressed as and and They represent the i-th beetle in the first beetle population, the i-th beetle in the second beetle population, and the i-th beetle in the third beetle population, respectively; Set the first maximum number of iterations to The second maximum number of iterations is And the third maximum number of iterations is The optimization search space dimensions of the first day beetle population, the second day beetle population and the third day beetle population are all set to be two-dimensional; S23422: According to the first initial weight and the first initial threshold Randomly generate the initial position set of the first day cattle population According to the second initial weight and the second initial threshold Randomly generate the initial position set of the cattle population on the second day The third initial position set of the Tianniu population is randomly generated according to the third initial weight l1 and the third initial threshold l2 n i ′1 and n i ′2 respectively represents the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the first longhorn beetle population; and Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the second longhorn beetle population; and Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th longhorn beetle in the third longhorn beetle population; The fitness functions of the first beetle population, the second beetle population, and the third beetle population are set as the first fitness function, the second fitness function, and the third fitness function, respectively; the function formulas of the first fitness function, the second fitness function, and the third fitness function are respectively as follows: Wherein, o1, o2 and o3 represent the first fitness function, the second fitness function and the third fitness function respectively; o1′, o′2 and o3′ represent the first bias, the second bias and the third bias respectively; S23423, start iterative operation, set the first current iteration number Second current iteration number and the third current iteration number In each round of iteration, the positions of the individual beetles in the first beetle population initial position set, the second beetle population initial position set, and the third beetle population initial position set are updated according to the function formulas of the first fitness function, the second fitness function, and the third fitness function; and in each round of iteration, the global best beetle individual and the global best fitness in the first beetle population, the second beetle population, and the third beetle population are obtained; S23424, when When the first beetle population stops the iteration operation, the optimal first weight value is output and the optimal first threshold when When , the second cattle population stops the iteration operation and outputs the optimal second weight and the optimal second threshold when When , the third beetle population stops the iterative operation and outputs the optimal third weight n1 and the optimal third threshold n2; S23425, the optimal first weight and the optimal first threshold Substitute the trained first BP neural network model to obtain the final first BP neural network model; replace the optimal second weight and the optimal second threshold Substitute the trained second BP neural network model to obtain the final second BP neural network model; substitute the optimal third weight n1 and the optimal third threshold n2 into the trained third BP neural network model to obtain the final third BP neural network model.
8. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 1 is characterized in that: The S3 comprises the following steps: S31, using the final first BP neural network model to predict the data of the first result data set matrix to be predicted at a future time, to obtain a first prediction data matrix p1; using the final second BP neural network model to predict the data of the second result data set matrix to be predicted at a future time, to obtain a second prediction data matrix p2; using the final third BP neural network model to predict the data of the third result data set matrix to be predicted at a future time, to obtain a third prediction data matrix p3; The first prediction data matrix p1, the second prediction data matrix p2 and the third prediction data matrix p3 are respectively as follows: Among them, p 1ij It indicates the perception data set of each intersection in the i-th test area at the j-th time point in the future. p 1ijk represents the predicted distance between the vehicle and the kth intersection when the vehicle perceives the kth intersection in the ith test area at the jth time point in the future; p 2ij It indicates the perception data set of each traffic light in the i-th test area at the j-th time point in the future. p 2ijk represents the predicted distance between the vehicle and the kth traffic light when the vehicle perceives the kth traffic light in the ith test area at the jth time point in the future; p 3ij It represents the perception dataset that predicts the obstacles of each road sub-section in the i-th test area at the j-th time point in the future. p 3ijk represents the predicted minimum distance between the vehicle and the obstacle when the vehicle perceives the obstacle in the i-th test area at the j-th time point in the future; p′ represents the total number of time points set for prediction in the future; S32, setting the sum of the time required for the perception module to transmit the intersection perception data, the traffic light perception data and the obstacle perception data to the decision module, the time required for the decision module to make a decision and the time required for the decision module to adjust the vehicle to be and They are recorded as the first decision-making time, the second decision-making time and the third decision-making time respectively; the current speed of the vehicle is set to S33, according to the first prediction data matrix p1, the second prediction data matrix p2, the third prediction data matrix p3, the first decision consumption time Second decision-making time Third decision-making time and the vehicle's current speed Determine whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle's driving speed.
9. The safety monitoring method based on big data vehicle-mounted intelligent driving system according to claim 8 is characterized in that: The S33 comprises the following steps: Otherwise, there is no need to issue an alert to the vehicle driving system; S332: After the vehicle driving system issues an alarm, the vehicle driving system needs to reduce the vehicle's driving speed so that and and 10. A system for implementing the safety monitoring method based on a big data vehicle-mounted intelligent driving system according to any one of claims 1 to 9, characterized in that: It includes a road intersection data acquisition module, a traffic light number data acquisition module, an obstacle data acquisition module, a first prediction module, a second prediction module and a final determination module; The road intersection data collection module is used to collect the number of road intersections in each area and obtain a first test result data set matrix; The traffic light number data collection module is used to collect the number of traffic lights in each area and obtain a second test result data set matrix; The obstacle data collection module is used to collect obstacle quantity data of each road sub-section in each area and obtain a third test result data set matrix; The first prediction module is used to predict the perceived efficiency at a future moment according to the first test result data set matrix, the second test result data set matrix, and the third test result data set matrix; The second prediction module is used to use the final first BP neural network model, the final second BP neural network model and the final third BP neural network model to respectively predict the intersection perception data, the traffic light perception data and the obstacle intersection perception data at the future moment; The final judgment module is used to determine whether it is necessary to issue an alarm to the vehicle driving system and adjust the vehicle driving speed based on the first prediction data matrix, the second prediction data matrix, the third prediction data matrix, the first decision consumption time, the second decision consumption time, the third decision consumption time and the current driving speed of the vehicle.
Citation Information
Patent Citations
A security monitoring method and system
CN112002102B
Driver information reminding device and system based on traffic volume forecast
CN108198447A
Intelligent obstacle avoidance system and method of autonomous vehicle
CN110371112A
Intrusion detection method for optimizing regularization extreme learning machine through improved longicorn swarm algorithm
CN111416797A
Intersection passing method based on NARX neural network and C-V2X technology
CN113450564A
