An intelligent planning and management system for ambulance emergency routes based on big data

Abstract

The present invention discloses an intelligent planning and management system for emergency routes of ambulances based on big data. The system extracts the traffic volume and speed information at each moment from the original traffic flow data, uses the historical trend algorithm and the exponential smoothing algorithm to repair the data missing due to the failure of the acquisition equipment, converts the processed data into time series data, inputs the time series data into a trained deep neural network for training to predict the vehicle information at the future moment, builds a driving path model of the ambulance according to the vehicle information and the network deployment of the area where the ambulance is located, completes the route matching of the ambulance, performs time-consuming analysis and evaluates the road condition of the road where the ambulance is located at the future moment according to the driving path model, uses the ant colony algorithm to optimize the road condition information to obtain the optimal path planning of the ambulance, provides a fast and accurate emergency route, improves the accuracy of the preset traffic volume, and improves the response rate of the ambulance.

Description

Technical Field

[0001] The present invention belongs to the technical field of route planning, and in particular relates to an intelligent planning and management system for ambulance emergency routes based on big data. Background Art

[0002] Smart transportation is the most important component of smart city construction. Traffic vehicle monitoring systems are deployed in various fields to build smart transportation. The existing traffic monitoring system not only realizes vehicle monitoring and tracking, but also can effectively detect a wealth of peripheral information such as vehicle speed and driving direction. Data mining based on traffic monitoring information has important application value and practical significance in the fields of traffic congestion prediction, urban planning and air pollution assessment. However, the deployment of medical networks in suburban areas will have limited network resources, sparse and unevenly distributed communication resources, and differences in computing resources, which can easily lead to difficulties in dispatching ambulances. Due to the uncertainty of traffic conditions, ambulances will spend a lot of time on the driving route, which will delay the emergency time. Therefore, it is necessary to provide an intelligent planning and management system for ambulance emergency routes that optimizes the driving route of ambulances and improves the rescue time. Summary of the invention

[0003] In view of this, the present invention provides an intelligent planning and management system for ambulance emergency routes based on big data, which can improve the accurate prediction of traffic flow and provide the optimal driving route, so as to solve the above-mentioned technical problems, and is specifically implemented by the following technical solutions.

[0004] The present invention provides an intelligent planning and management system for ambulance emergency routes based on big data, comprising:

[0005] The data preprocessing unit is used to extract the traffic volume and speed information at each moment from the original traffic flow data, and to use the historical trend algorithm and exponential smoothing algorithm to repair the missing data due to the failure of the acquisition equipment, and to convert the processed data into time series data as the input of the evaluation parameter prediction part;

[0006] An evaluation parameter prediction unit, used for inputting time series data into a trained deep neural network for training to predict vehicle information at future moments;

[0007] A path construction unit, used to construct a driving path model of the ambulance according to the vehicle information and the network deployment in the area where the ambulance is located and complete the route matching of the ambulance;

[0008] A road condition evaluation module is used to perform time consumption analysis based on the driving path model and evaluate the road condition of the ambulance at a future time to obtain road condition information;

[0009] The planning and management module is used to optimize the road condition information using the ant colony algorithm to obtain the optimal path planning for the ambulance and complete the intelligent planning and management of the ambulance emergency route.

[0010] As a preferred embodiment of the above technical solution, the time series data is input into a trained deep neural network for training to predict vehicle information at future moments, including:

[0011] The Spark cluster is used to train the prediction model and predict the traffic flow and speed information at future times. The average speed, road section saturation and traffic flow density at future times are calculated based on the prediction results. The road section saturation refers to the ratio of the actual traffic flow of the road section to the maximum traffic capacity to reflect the actual load capacity of the road. The calculation expression of the road section saturation is: Among them, V represents the current traffic volume of the road, and C represents the maximum traffic capacity of the road;

[0012] Traffic flow density refers to the number of vehicles per unit length of the road in unit time, which reflects the density of vehicles on the road. The calculation expression of traffic flow density is: Among them, f represents the traffic flow and v represents the average speed.

[0013] As a preferred embodiment of the above technical solution, a driving path model of an ambulance is constructed according to vehicle information and the network deployment in the area where the ambulance is located, and the route matching of the ambulance is completed, including:

[0014] The traffic network for ambulance route planning is an irregular non-periodic dynamic directed graph denoted as Z = (G, T), where G represents a static subnet and T represents a period. G changes as T changes. In the period, it is composed of traffic networks, then G = (V, E), where V = {v 1 ,v 2 ...v n} represents the vertex set, E={e 1 ,e 2 ...e m} is an edge set, and the snapshots of the road network at different periods are obtained according to the time-varying traffic road network {G 1 , G 2 ...G T}, L represents the path of the ambulance;

[0015] Establish a trajectory model based on the snapshots and adjust the trajectory model parameters λ(π, A, B) so that the trajectory data in the training data matches the correct path under the adjusted parameter model;

[0016] The vehicle's trajectory information and the corresponding route information are input, and the Viterbi algorithm and the existing observation sequence are used to determine the possible hidden sequences, that is, the actual location points corresponding to the observed GPS trajectory points, and the probability of the corresponding sequence is calculated according to the parameter matrix of the HMM model. The hidden sequence with the maximum probability, that is, the actual running trajectory of the vehicle, is found to complete the matching with the route.

[0017] As a preferred embodiment of the above technical solution, the error in GPS measurement can be described as p i With c ij The normal distribution of the distance between 2 ), using a zero-mean normal distribution based on empirical evaluation, the observation probability is the distance factor, and the expression for the observation probability is Where σ represents the standard deviation of GPS error, which is 20 meters, dist(p i , c ij ) represents the distance from the trajectory point to the candidate point;

[0018] Among them, the current point p i The precision p i ·Draw a circle with hdop as radius, and select the set of covered sections as candidate sections. The candidate sections are composed of e ij Indicates p i The candidate road segments are calculated i To e j The set of nearest points, when p i In the candidate section j The projection point on is at e j When the candidate point is on, it is the projection point; if the candidate road segment e j The projection point on is not on e j If , then the point on the road segment with the shortest distance to this point is the corresponding candidate point.

[0019] As a preferred embodiment of the above technical solution, the transition probability of the HMM model is used to describe the process from one state to another. Let c i-1 and c i Represent two adjacent trajectory points p i-1 and p i The candidate point of c i-1 For p i-1 When the matching point is p i-1 to p i The real path is c i-1 to c i The possibility of, the corresponding expression is Among them, path(c i-1 , c i ) represents the path candidate point c i-1 to c iThe shortest path, λ represents the probability coefficient;

[0020] By adjusting the model parameters λ(π, A, B), the GPS trajectory points observed in the data set can obtain the hidden state, that is, the actual position sequence of the vehicle, with the maximum probability, including:

[0021] Step 1: Input the initial state transfer matrix A, observation probability matrix B, initial state π, observation sequence O, maximum number of iterations m, error σ, then the corresponding parameters are set to

[0022]

[0023] Step 2: Calculate the model under k iterations The corresponding parameter of the model is γ t (i) and ξ t (i, j) represents the estimated value corresponding to the last iteration;

[0024] Step 3: Calculate and observe Xu Lei in the model The output probability P(O|λ k ), if the output probability |P(O|λ k )-P(O|λ k-1 )|≥ξ, then return to the second step to continue the iteration, otherwise the model λ will be re-estimated k as the final model.

[0025] As a preferred embodiment of the above technical solution, the time consumption analysis and the road condition of the road where the ambulance is located at a future time are evaluated according to the driving path model to obtain the road condition information, including:

[0026] For each road section, the range of network quality on the road section where the ambulance is traveling is calculated based on the current network communication speed, path loss, and shadow fading. The default value r(u,o) represents the network communication rate between the ambulance u and the base station o. The corresponding expression is: Among them, b u represents the bandwidth allocated by the base station to ambulance u, b u max , h represents the channel gain, p re Indicates the signal receiving power, N 0 represents the average noise power, p tra represents the signal transmission power, e represents the radiation power, represents the polarization loss factor, c represents the detection wavelength, f represents the frequency, D(·) represents the direction of the antenna, θ, Two parameters representing polar coordinates;

[0027] ​When the ambulance is in the shadow area, the signal field strength is weak; when the ambulance passes through the shadow area, the signal field strength is strong. In the propagation link of mobile communications, radio waves will collide with various obstacles to produce radio wave shadow areas.

[0028] As a preferred embodiment of the above technical solution, shadow fading obeys log-normal distribution, and the preset Pathss represents the path loss between the ambulance u and the base station o, and p(ψ) represents the expression of the log-normal shadow model: Where, ζ=10 / ln10, μ ψdB Denotes ψ in dB dB The mean value, σ ψdB ψ means ψ dB The standard deviation of

[0029] The superposition of path loss and shadow fading models can reflect both signal propagation and path loss, where the power decreases with the increase of propagation distance and the path loss causes random attenuation as the shadow changes;

[0030] The superposition of the models is based on the path loss model Pathss ψdB and the shadow fading model with a mean value of 0dB, the received power p re and the transmission power p tra The expression is

[0031] The expression is a combination of path loss and shadow fading, p(ψ) means the mean is 0, and the variance is The path loss decreases linearly with 10lgd. 0 represents the reference distance of the antenna far field, and γ represents the path loss exponent.

[0032] As a preferred embodiment of the above technical solution, the processed data is converted into time series data as input of the evaluation parameter prediction part, including:

[0033] The model input expression is

[0034] in, represents the traffic volume to be predicted, i represents the day, j represents the time period, and N represents the total number of samples. It represents the traffic volume in the jth period of the i-th day, reflecting the horizontal periodicity of the data; It means that the traffic flow in the jth period of the ith day in the previous n weeks, that is, the traffic flow at a certain moment should be affected by the traffic flow at the same moment in the previous n weeks, reflecting the longitudinal periodicity of the data. m and n are positive integers.

[0035] As a preferred embodiment of the above technical solution, an ant colony algorithm is used to optimize the road condition information to obtain the optimal path planning for the ambulance, including:

[0036] Road condition evaluation is to analyze the road condition at the future moment through fuzzy comprehensive evaluation. The weight of each indicator in the fuzzy comprehensive evaluation algorithm is determined by the entropy method. First, the factor set U = {u 1 ,u 2 ,K,u m} and evaluation set V = {v 1 ,v 2 ...v m};

[0037] Determine the weight coefficient of each indicator in the factor set W = {w 1 ,w 2 ,K,w m}, all weights are always positive and the sum of all weights is 1; the evaluation matrix R is constructed using the membership function;

[0038] The evaluation result is obtained by synthesizing W and R using the fuzzy synthesis operator.

[0039] As a preferred embodiment of the above technical solution, the road conditions are divided into six levels: extremely smooth, smooth, light congestion, medium congestion, heavy congestion and locked. The result of the preset evaluation parameter prediction part is R = (R 1 ,R 2 ,L,R i ) T , select the average vehicle speed, road section saturation and traffic flow density as the road condition evaluation parameters, then i = 3, R i Represents the traffic volume, road section saturation and traffic flow density corresponding to a certain time interval, then the expression of the evaluation matrix is: For each factor i, the membership degree r of the jth level is obtained through the membership function ij ;

[0040] The descending semi-trapezoidal membership function and the ascending semi-trapezoidal membership function are used for very smooth traffic and traffic jam respectively. The horizontal axis of the membership function represents the value of each factor index of each traffic sample. 1 ,u 2 ,L,u 5} represents the threshold range of each factor indicator, {k 1 ,k 2 ,L,k 10} represents the linear value of the adjacent threshold of each factor indicator, and the vertical axis 1 represents the desired congestion state, and 0 represents the non-congestion state;

[0041] According to the positive factor indicators of vehicle flow, traffic density, and road saturation, the function is from very smooth to locked from left to right. According to the trapezoidal membership function r ij The calculation expression is and Where x represents r ij The entropy method is used to obtain the weight coefficients of the three indicators of average vehicle speed, road section saturation and traffic flow density w = {w 1 ,w 2 ,w 3}.

[0042] The present invention provides an intelligent planning and management system for emergency routes of ambulances based on big data. The system extracts the traffic volume and speed information at each moment from the original traffic flow data, and uses the historical trend algorithm and the exponential smoothing algorithm to repair the data missing due to the failure of the acquisition equipment. The processed data is converted into time series data as the input of the evaluation parameter prediction part. The time series data is input into the trained deep neural network for training to predict the vehicle information at the future moment. The driving path model of the ambulance is constructed according to the vehicle information and the network deployment of the area where the ambulance is located, and the route matching of the ambulance is completed. The time-consuming analysis and evaluation of the road conditions of the road where the ambulance is located at the future moment are performed according to the driving path model to obtain the road condition information. The ant colony algorithm is used to optimize the road condition information to obtain the optimal path planning of the ambulance, and the intelligent planning and management of the emergency route of the ambulance is completed. The system can provide a fast and accurate emergency route, effectively predict the current traffic conditions, improve the accuracy of the preset traffic volume, and improve the response rate of the ambulance. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 This is a structural block diagram of the intelligent planning and management system for ambulance emergency routes based on big data of the present invention. DETAILED DESCRIPTION

[0045] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0046] See also Figure 1 The present invention provides an intelligent planning and management system for emergency ambulance routes based on big data, including:

[0047] The data preprocessing unit is used to extract the traffic volume and speed information at each moment from the original traffic flow data, and to use the historical trend algorithm and exponential smoothing algorithm to repair the missing data due to the failure of the acquisition equipment, and to convert the processed data into time series data as the input of the evaluation parameter prediction part;

[0048] An evaluation parameter prediction unit, used for inputting time series data into a trained deep neural network for training to predict vehicle information at future moments;

[0049] A path construction unit, used to construct a driving path model of the ambulance according to the vehicle information and the network deployment in the area where the ambulance is located and complete the route matching of the ambulance;

[0050] A road condition evaluation module is used to perform time consumption analysis based on the driving path model and evaluate the road condition of the ambulance at a future time to obtain road condition information;

[0051] The planning and management module is used to optimize the road condition information using the ant colony algorithm to obtain the optimal path planning for the ambulance and complete the intelligent planning and management of the ambulance emergency route.

[0052] As a preferred embodiment of the above technical solution, the time series data is input into a trained deep neural network for training to predict vehicle information at future moments, including:

[0053] The Spark cluster is used to train the prediction model and predict the traffic flow and speed information at future times. The average speed, road section saturation and traffic flow density at future times are calculated based on the prediction results. The road section saturation refers to the ratio of the actual traffic flow of the road section to the maximum traffic capacity to reflect the actual load capacity of the road. The calculation expression of the road section saturation is: Among them, V represents the current traffic volume of the road, and C represents the maximum traffic capacity of the road;

[0054] Traffic flow density refers to the number of vehicles per unit length of the road in unit time, which reflects the density of vehicles on the road. The calculation expression of traffic flow density is: Among them, f represents the traffic flow and v represents the average speed.

[0055] In this embodiment, the ambulance driving path model is constructed according to the vehicle information and the network deployment of the area where the ambulance is located, and the ambulance route matching is completed, including: the traffic network for the ambulance route planning is preset as an irregular non-periodic dynamic directed graph denoted as Z = (G, T), where G represents a static subnet, T represents a period, G changes as T changes, and is composed of a traffic network within the period, then G = (V, E), where V = {v 1 ,v 2 ...v n} represents the vertex set, E={e 1 ,e 2 ...e m} is an edge set, and the snapshots of the road network at different periods are obtained according to the time-varying traffic road network {G 1 ,G 2 ...G T}, L represents the driving path of the ambulance; a trajectory model is established based on the snapshots and the trajectory model parameters λ(π,A,B) are adjusted so that the trajectory data in the training data can be matched to the correct path under the adjusted parameter model; the vehicle trajectory information and the corresponding route information are input, and the Viterbi algorithm and the existing observation sequence are used to determine the possible hidden sequence, that is, the actual location points corresponding to the observed GPS trajectory points, and the probability of the corresponding sequence is calculated according to the parameter matrix of the HMM model, and the hidden sequence with the maximum probability, that is, the actual running trajectory of the vehicle, is found to complete the matching with the route.

[0056] It should be noted that the error in GPS measurement can be described as p i With c ij The normal distribution of the distance between 2 ), using a zero-mean normal distribution based on empirical evaluation, the observation probability is the distance factor, and the expression for the observation probability is Where σ represents the standard deviation of GPS error, which is 20 meters, dist(p i ,c ij ) represents the distance from the trajectory point to the candidate point; where the current point p i The precision p i ·Draw a circle with hdop as radius, and select the set of covered sections as candidate sections. The candidate sections are composed of e ij Indicates p i The candidate road segments are calculated i To e j The set of nearest points, when p i In the candidate section j The projection point on is at e j When the candidate point is on, it is the projection point; if the candidate road segment e j The projection point on is not on e j If , then the point on the road segment with the shortest distance to this point is the corresponding candidate point.

[0057] It should be understood that if the overall trajectory of the moving object is used as a trajectory object for route matching, the intercepted road section that can match the trajectory will be lost. The route direction corresponding to the neighboring points of some trajectory points changes too quickly, and there may be too few trajectory point sequences after being intercepted. At this time, more trajectory points need to be intercepted to retain the characteristics of the vehicle trajectory. After intercepting the route and trajectory sequence, it is also necessary to select appropriate candidate sections and candidate points. The candidate point is the closest point on the candidate section to which the latest point corresponds. The purpose of introducing HMM is to find the optimal matching route. Then it is necessary to introduce the Viterbi algorithm to find the maximum probability path to solve the hidden Markov prediction problem. The Viterbi algorithm gradually solves the joint probability of the state based on the observation probability and the transition probability, and records the previous state of each hidden state, and finally obtains the end point of the optimal path, and traces back to the entire optimal path based on the previous state information that has been recorded. The Viterbi algorithm infers the hidden state from the observed sampling points and calculates the shortest path for each continuous candidate pair. It is a dynamic programming algorithm for generating the most likely path sequence. The principle of road condition prediction is to predict the trend of future road conditions based on the evaluation parameters of past, present and future traffic conditions in the traffic flow. By extracting the traffic volume and speed information at each moment from the original traffic flow data, and using the historical trend algorithm and exponential smoothing algorithm to repair the data missing due to the failure of the acquisition equipment, the processed data is converted into time series data as the input of the evaluation parameter prediction part, and the time series data is input into the trained deep neural network for training to predict the vehicle information at future moments. According to the vehicle information and the network deployment of the area where the ambulance is located, the driving path model of the ambulance is constructed and the route matching of the ambulance is completed. According to the driving path model, the time consumption analysis and evaluation of the road conditions of the ambulance at future moments are carried out to obtain the road condition information. The ant colony algorithm is used to optimize the road condition information to obtain the optimal path planning of the ambulance, and the intelligent planning and management of the ambulance emergency route are completed. It can provide fast and accurate emergency routes, effectively predict the current traffic conditions, improve the accuracy of the preset traffic flow, and improve the response rate of the ambulance.

[0058] Optionally, the transition probability of the HMM model is used to describe the process from one state to another. Let c i-1 and c i Represent two adjacent trajectory points p i-1 and p i The candidate point of c i-1 For p i-1 When the matching point is p i-1 to p i The real path is c i-1 to c i The possibility of, the corresponding expression is Among them, path(c i-1 ,ci ) represents the path candidate point c i-1 to c i The shortest path, λ represents the probability coefficient;

[0059] By adjusting the model parameters λ(π,A,B), the GPS trajectory points observed in the data set can obtain the hidden state, that is, the actual position sequence of the vehicle, with the maximum probability, including:

[0060] Step 1: Input the initial state transfer matrix A, observation probability matrix B, initial state π, observation sequence O, maximum number of iterations m, error σ, then the corresponding parameters are set to

[0061]

[0062] Step 2: Calculate the model under k iterations The corresponding parameter of the model is γ t (i) and ξ t (i, j) represents the estimated value corresponding to the last iteration;

[0063] Step 3: Calculate and observe Xu Lei in the model The output probability P(O|λ k ), if the output probability |P(O|λ k )-P(O|λ k-1 )|≥ξ, then return to the second step to continue the iteration, otherwise the model λ will be re-estimated k as the final model.

[0064] In this embodiment, the road condition information is obtained by performing time-consuming analysis and evaluating the road condition of the ambulance at the future time according to the driving path model, including: calculating the range of network quality on the road section where the ambulance is traveling according to the current network communication speed, path loss and shadow fading for each road section, and the preset r(u,o) represents the network communication rate between the ambulance u and the base station o, and the corresponding expression is: Among them, b u represents the bandwidth allocated by the base station to ambulance u, b u max , h represents the channel gain, p re Indicates the signal receiving power, N 0 represents the average noise power, p tra represents the signal transmission power, e represents the radiation power, represents the polarization loss factor, c represents the detection wavelength, f represents the frequency, D(·) represents the direction of the antenna, θ, ​Represents two parameters of polar coordinates; when the ambulance is in the shadow area, the signal field strength is weak; when the ambulance passes through the shadow area, the signal field strength is strong. In the propagation link of mobile communications, radio waves will collide with various obstacles, resulting in radio wave shadow areas.

[0065] It should be noted that shadow fading obeys the log-normal distribution. Pathss is preset to represent the path loss between the ambulance u and the base station o. p(ψ) represents the expression of the log-normal shadow model: Where, ζ=10 / ln10, μ ψdB Denotes ψ in dB dB The mean value, σ ψdB ψ means ψ dB The standard deviation of

[0066] The superposition of path loss and shadow fading models can reflect both signal propagation and path loss, where the power decreases with the increase of propagation distance and the path loss causes random attenuation as the shadow changes;

[0067] The superposition of the models is based on the path loss model Pathss ψdB and the shadow fading model with a mean value of 0dB, the received power p re and the transmission power p tra The expression is

[0068] The expression is a combination of path loss and shadow fading, p(ψ) means the mean is 0, and the variance is The path loss decreases linearly with 10lgd. 0 represents the reference distance of the antenna far field, and γ represents the path loss exponent, thereby improving the communication data transmission efficiency of the ambulance.

[0069] Optionally, the processed data is converted into time series data as input for the evaluation parameter prediction part, including:

[0070] The model input expression is

[0071] in, represents the traffic volume to be predicted, i represents the day, j represents the time period, and N represents the total number of samples. It represents the traffic volume in the jth period of the i-th day, reflecting the horizontal periodicity of the data; It means that the traffic flow in the jth period of the ith day in the previous n weeks, that is, the traffic flow at a certain moment should be affected by the traffic flow at the same moment in the previous n weeks, reflecting the longitudinal periodicity of the data. m and n are positive integers.

[0072] In this embodiment, an ant colony algorithm is used to optimize the road condition information to obtain the optimal path planning for the ambulance, including:

[0073] Road condition evaluation is to analyze the road condition at the future moment through fuzzy comprehensive evaluation. The weight of each indicator in the fuzzy comprehensive evaluation algorithm is determined by the entropy method. First, the factor set U = {u 1 ,u 2 ,K,u m} and evaluation set V = {v 1 ,v 2 ...v m};

[0074] Determine the weight coefficient of each indicator in the factor set W = {w 1 ,w 2 ,K,w m}, all weights are always positive and the sum of all weights is 1; the evaluation matrix R is constructed using the membership function;

[0075] The evaluation result is obtained by synthesizing W and R using the fuzzy synthesis operator.

[0076] It should be noted that the traffic conditions are divided into six levels: extremely smooth, smooth, light congestion, medium congestion, heavy congestion and deadlock. The result of the preset evaluation parameter prediction part is R = (R 1 ,R 2 ,L,R i ) T , select the average vehicle speed, road section saturation and traffic flow density as the road condition evaluation parameters, then i = 3, R i Represents the traffic volume, road section saturation and traffic flow density corresponding to a certain time interval, then the expression of the evaluation matrix is: For each factor i, the membership degree r of the jth level is obtained through the membership function ij ;

[0077] The descending semi-trapezoidal membership function and the ascending semi-trapezoidal membership function are used for very smooth traffic and traffic jam respectively. The horizontal axis of the membership function represents the value of each factor index of each traffic sample. 1 ,u 2 ,L,u 5} represents the threshold range of each factor indicator, {k 1 ,k 2 ,L,k 10} represents the linear value of the adjacent threshold of each factor indicator, and the vertical axis 1 represents the desired congestion state, and 0 represents the non-congestion state;

[0078] According to the positive factor indicators of vehicle flow, traffic density, and road saturation, the function is from very smooth to locked from left to right. According to the trapezoidal membership function rij The calculation expression is and Where x represents r ij The entropy method is used to obtain the weight coefficients of the three indicators of average vehicle speed, road section saturation and traffic flow density w = {w 1 ,w 2 ,w 3}, effectively predicting and analyzing traffic conditions and optimizing the intelligent planning of ambulance emergency routes.

[0079] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0080] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. An intelligent planning and management system for ambulance emergency routes based on big data, characterized in that: include: The data preprocessing unit is used to extract the traffic volume and speed information at each moment from the original traffic flow data, and to use the historical trend algorithm and exponential smoothing algorithm to repair the missing data due to the failure of the acquisition equipment, and to convert the processed data into time series data as the input of the evaluation parameter prediction part; An evaluation parameter prediction unit, used for inputting time series data into a trained deep neural network for training to predict vehicle information at future moments; A path construction unit, used to construct a driving path model of the ambulance according to the vehicle information and the network deployment in the area where the ambulance is located and complete the route matching of the ambulance; A road condition evaluation module is used to perform time consumption analysis based on the driving path model and evaluate the road condition of the ambulance at a future time to obtain road condition information; The planning and management module is used to optimize the road condition information using the ant colony algorithm to obtain the optimal path planning for the ambulance and complete the intelligent planning and management of the ambulance emergency route; Input time series data into the trained deep neural network for training to predict vehicle information at future times, including: The Spark cluster is used to train the prediction model and predict the traffic flow and speed information at future times. The average speed, road section saturation and traffic flow density at future times are calculated based on the prediction results. The road section saturation refers to the ratio of the actual traffic flow of the road section to the maximum traffic capacity to reflect the actual load capacity of the road. The calculation expression of the road section saturation is: Among them, V represents the current traffic volume of the road, and C represents the maximum traffic capacity of the road; Traffic flow density refers to the number of vehicles per unit length of the road in unit time, which reflects the density of vehicles on the road. The calculation expression of traffic flow density is: Among them, f represents the traffic volume, v represents the average speed; Based on the vehicle information and the network deployment in the area where the ambulance is located, the ambulance's driving path model is constructed and the ambulance's route matching is completed, including: The traffic network for ambulance route planning is an irregular non-periodic dynamic directed graph denoted by Z = (G, T), where G represents a static subnet and T represents a period. G changes with T and is composed of traffic networks within the period. Then G = (V, E), where V = {v1, v2...v n } represents the vertex set, E={e1,e2...e m } is an edge set, and the road network snapshots {G1,G2...G T }, L represents the path of the ambulance; Establish a trajectory model based on the snapshots and adjust the trajectory model parameters λ(π, A, B) so that the trajectory data in the training data matches the correct path under the adjusted parameter model; Input the vehicle's trajectory information and the corresponding route information, use the Viterbi algorithm and the existing observation sequence to determine the possible hidden sequence, that is, the actual location point corresponding to the observed GPS trajectory point, and calculate the probability of the corresponding sequence based on the parameter matrix of the HMM model, find the hidden sequence with the maximum probability, that is, the actual running trajectory of the vehicle, and complete the matching with the route; The time consumption is analyzed based on the driving path model and the road condition of the ambulance at the future time is evaluated to obtain the road condition information, including: For each road section, the range of network quality on the road section where the ambulance is traveling is calculated based on the current network communication speed, path loss, and shadow fading. The default value r(u,o) represents the network communication rate between the ambulance u and the base station o. The corresponding expression is: Among them, b u represents the bandwidth allocated by the base station to ambulance u, b u max , h represents the channel gain, p re represents the received signal power, N0 represents the average noise power, and p tra represents the signal transmission power, e represents the radiation power, represents the polarization loss factor, c represents the detection wavelength, f represents the frequency, D(·) represents the direction of the antenna, θ, Two parameters representing polar coordinates;​ When the ambulance is in the shadow area, the signal field strength is weak; when the ambulance passes through the shadow area, the signal field strength is strong. In the propagation link of mobile communications, radio waves will collide with various obstacles, resulting in radio wave shadow areas; The ant colony algorithm is used to optimize the road condition information to obtain the optimal path planning for the ambulance, including: Road condition evaluation is to analyze the road condition at the future moment through fuzzy comprehensive evaluation. The weight of each indicator in the fuzzy comprehensive evaluation algorithm is determined by the entropy method. First, the factor set U = {u1,u2,K,u m } and evaluation set V = {v1,v2...v m }; Determine the weight coefficient of each indicator in the factor set W = {w1,w2,K,w m }, all weights are always positive and the sum of all weights is 1; the evaluation matrix R is constructed using the membership function; The evaluation result is obtained by synthesizing W and R using the fuzzy synthesis operator.

2. The intelligent planning and management system for emergency ambulance routes based on big data according to claim 1 is characterized in that: Also includes: The error in GPS measurement can be described as p i With c ij The normal distribution of the distance between 2 ), using a zero-mean normal distribution based on empirical evaluation, the observation probability is the distance factor, and the expression for the observation probability is Where σ represents the standard deviation of GPS error, which is 20 meters, dist(p i ,c ij ) represents the distance from the trajectory point to the candidate point; Among them, the current point p i The precision p i ·Draw a circle with hdop as radius, and select the set of covered sections as candidate sections. The candidate sections are composed of e ij Indicates p i The candidate road segments are calculated i To e j The set of nearest points, when p i In the candidate section j The projection point on is at e j When the candidate point is on, it is the projection point; if the candidate road segment e j The projection point on is not on e j If , then the point on the road segment with the shortest distance to this point is the corresponding candidate point.

3. The intelligent planning and management system for emergency ambulance routes based on big data according to claim 2 is characterized in that: The transition probability of the HMM model is used to describe the process from one state to another. Let c i-1 and c i Represent two adjacent trajectory points p i-1 and p i The candidate point of c i-1 For p i-1 When the matching point is p i-1 to p i The real path is c i-1 to c i The possibility of, the corresponding expression is Among them, path(c i-1 ,c i ) represents the path candidate point c i-1 to c i The shortest path, λ represents the probability coefficient; By adjusting the model parameters λ(π,A,B), the GPS trajectory points observed in the data set can obtain the hidden state, that is, the actual position sequence of the vehicle, with the maximum probability, including: Step 1: Input the initial state transfer matrix A, observation probability matrix B, initial state π, observation sequence O, maximum number of iterations m, error σ, then the corresponding parameters are set to Step 2: Calculate the model under k iterations The corresponding parameter of the model is γ t (i) and ξ t (i, j) represents the estimated value corresponding to the last iteration; Step 3: Calculate and observe Xu Lei in the model The output probability P(O|λ k ), if the output probability |P(O|λ k )-P(O|λ k-1 )|≥ξ, then return to the second step to continue the iteration, otherwise the model λ will be re-estimated k as the final model.

4. The intelligent planning and management system for emergency ambulance routes based on big data according to claim 1 is characterized in that: Shadow fading obeys the log-normal distribution. Pathss is assumed to represent the path loss between the ambulance u and the base station o. p(ψ) represents the expression of the log-normal shadow model: Where, ζ=10 / ln10, μ ψdB Denotes ψ in dB dB The mean value, σ ψdB ψ means ψ dB The standard deviation of The superposition of path loss and shadow fading models can reflect both signal propagation and path loss, where the power decreases with the increase of propagation distance and the path loss causes random attenuation as the shadow changes; The superposition of the models is based on the path loss model Pathss ψdB and the shadow fading model with a mean value of 0dB, the received power p re and the transmission power p tra The expression is The expression is a combination of path loss and shadow fading, p(ψ) means the mean is 0, and the variance is is a Gaussian random variable, the path loss decreases linearly with 10lgd, d0 represents the reference distance of the antenna far field, and γ represents the path loss exponent.

5. The intelligent planning and management system for emergency ambulance routes based on big data according to claim 1 is characterized in that: The processed data is converted into time series data as input for the evaluation parameter prediction part, including: The model input expression is in, represents the traffic volume to be predicted, i represents the day, j represents the time period, and N represents the total number of samples. It represents the traffic volume in the jth period of the i-th day, reflecting the horizontal periodicity of the data; It means that the traffic flow in the jth period of the ith day in the previous n weeks, that is, the traffic flow at a certain moment should be affected by the traffic flow at the same moment in the previous n weeks, reflecting the longitudinal periodicity of the data. m and n are positive integers.

6. The intelligent planning and management system for emergency ambulance routes based on big data according to claim 1 is characterized in that: Also includes: The traffic conditions are divided into six levels: extremely smooth, smooth, light congestion, medium congestion, heavy congestion and deadlock. The result of the preset evaluation parameter prediction part is R = (R1, R2, L, R i ) T , select the average vehicle speed, road section saturation and traffic flow density as the road condition evaluation parameters, then i = 3, R i Represents the traffic volume, road section saturation and traffic flow density corresponding to a certain time interval, then the expression of the evaluation matrix is: For each factor i, the membership degree r of the jth level is obtained through the membership function ij ; The descending semi-trapezoidal membership function and the ascending semi-trapezoidal membership function are used for very smooth traffic and locked traffic respectively. The horizontal axis of the membership function represents the value of each factor index of each traffic sample, {u1,u2,L,u5} represents the threshold range of each factor index, {k1,k2,L,k 10 } represents the linear value of the adjacent threshold of each factor indicator, and the vertical axis 1 represents the desired congestion state, and 0 represents the non-congestion state; According to the positive factor indicators of vehicle flow, traffic density, and road saturation, the function is from very smooth to locked from left to right. According to the trapezoidal membership function r ij The calculation expression is and Where x represents r ij The entropy method is used to obtain the sample values ​​and the weight coefficients w = {w1, w2, w3} of the three indicators: average vehicle speed, road section saturation and traffic flow density.

Images