Intelligent subway monitoring system

Abstract

The present application relates to subway monitoring, in particular to a kind of intelligent subway monitoring system, vehicle terminal, vehicle data are collected by sensor module, simultaneously by positioning module, the position information corresponding to the vehicle data obtained by sensor module is collected, passenger distribution coefficient calculation module is used to calculate the passenger distribution coefficient of vehicle according to vehicle data, and passenger distribution coefficient, vehicle data and its corresponding position information are encapsulated and sent to ground base station;Control terminal, receive the data transmitted by ground base station, carry out data processing to the received data by data processing module, utilize control decision module to evaluate passenger flow and vehicle operation according to the received data after processing, and formulate control scheduling strategy, control scheduling strategy is sent to vehicle terminal by ground base station;The technical scheme provided by the present application can effectively overcome the defects that subway vehicle depot cannot be accurately and efficiently controlled and scheduled in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to subway monitoring, in particular to a smart subway monitoring system. BACKGROUND

[0002] With the rapid increase of urban population, the traffic congestion problem of large cities follows, the original basic transportation facilities have been unable to meet the current travel demand, and the main way to solve the problem at present is to build underground railway facilities. Subway can solve the problem of insufficient traditional public transport capacity, effectively alleviate traffic pressure, and has been widely promoted and constructed in many cities, which is also an important solution to solve the traffic congestion problem of large cities in China.

[0003] At present, the automation degree of subway control and dispatch is low, and the dispatchers of each post in the base mainly rely on manual command and dispatch, control and dispatch plan manual preparation and oral communication. Various systems are scattered, and the linkage between systems is limited, so it is difficult to accurately and efficiently control and dispatch the subway vehicle depot. SUMMARY

[0004] (I) Technical problems solved

[0005] In view of the above shortcomings of the prior art, the present application provides a smart subway monitoring system, which can effectively overcome the defects of the prior art that it is difficult to accurately and efficiently control and dispatch the subway vehicle depot.

[0006] (II) Technical solutions

[0007] In order to achieve the above purpose, the present application is realized by the following technical solutions:

[0008] A smart subway monitoring system comprises a vehicle-mounted terminal and a control terminal.

[0009] The vehicle-mounted terminal collects vehicle data through a sensor module, collects corresponding position information of the vehicle data obtained by the sensor module through a positioning module, calculates the passenger distribution coefficient of the vehicle according to the vehicle data by using a passenger distribution coefficient calculation module, and sends the passenger distribution coefficient, vehicle data and corresponding position information to the ground base station after encapsulation.

[0010] The control terminal receives the data transmitted by the ground base station, processes the received data through a data processing module, evaluates the passenger flow and vehicle operation according to the processed received data by using a control decision module, and formulates a control and dispatch strategy, which is sent to the vehicle-mounted terminal through the ground base station.

[0011] The data processing module fills in missing values in the passenger distribution coefficient and the vehicle data by using an improved non-negative matrix factorization (NMF), the improved NMF introduces a matrix representing the data structure of the missing values to improve the accuracy of the filled data, while maintaining the consistency of the filled data with the original data.

[0012] Preferably, the vehicle terminal collects vehicle data through the sensor module, and collects corresponding position information of the sensor module when collecting the vehicle data through the positioning module, including:

[0013] The vehicle terminal collects vehicle data through the sensor module installed in each carriage, and the vehicle data includes the area of the heat source region of each carriage collected by the infrared sensor, the number of heat source targets, and the running state data of each carriage collected by other sensors.

[0014] Meanwhile, the vehicle terminal collects corresponding position information of the sensor module when collecting the vehicle data through the positioning module installed in each carriage, and the position information includes the carriage number, the nearest station information and the GPS position information.

[0015] Preferably, the vehicle terminal calculates the passenger distribution coefficient of the vehicle according to the vehicle data by using the passenger distribution coefficient calculation module, including:

[0016] The vehicle terminal calculates the passenger distribution coefficient of the vehicle according to the area of the heat source region and the number of heat source targets of each carriage by using the passenger distribution coefficient calculation module, and uses the following formula:

[0017]

[0018] wherein PDC i is the passenger distribution coefficient of the vehicle i, S ij , S' ij are the area of the heat source region and the maximum area of the accommodated passengers of the carriage j in the vehicle i, N ij , N' ij are the number of heat source targets and the maximum number of accommodated passengers of the carriage j in the vehicle i, ω j is the weight coefficient of the carriage j in the vehicle i, which is set according to the functional attributes of the surrounding building facilities of the vehicle arrival station, j∈[1,n], and n is the number of carriages of the vehicle i.

[0019] Preferably, the vehicle terminal encapsulates the passenger distribution coefficient, the vehicle data and the corresponding position information and sends them to the ground base station, including:

[0020] After encapsulating the passenger distribution coefficient, the vehicle data and the corresponding position information, the vehicle terminal sends them to the ground base station through the wireless communication module.

[0021] After receiving the encapsulated data, the ground base station decodes and verifies it to ensure the accuracy and integrity of the data, and then transmits this data to the control terminal.

[0022] Preferably, the control terminal receives data transmitted from the ground base station and processes the received data through a data processing module, including:

[0023] The control terminal receives data transmitted from the ground base station and preprocesses the received data through the data processing module;

[0024] The data processing module uses an improved nonnegative matrix factorization (NMF) to fill in the missing values ​​in the preprocessed passenger distribution coefficients and vehicle data. The improved NMF introduces a matrix that represents the data structure of the missing values ​​to improve the accuracy of the data filling while maintaining the consistency between the filled data and the original data.

[0025] The data processing module preprocesses the received data, including removing outliers and smoothing the data.

[0026] Preferably, the data processing module uses an improved nonnegative matrix factorization (NMF) to fill in missing values ​​in the preprocessed passenger distribution coefficients and vehicle data, including:

[0027] S1. Organize the preprocessed passenger distribution coefficients and vehicle data into a non-negative matrix V;

[0028] In this matrix, missing values ​​in the non-negative matrix V are represented by 0. Each row of the matrix represents a time point, and each column of the matrix represents a data category. The data categories include the area of ​​the heat source region in each carriage, the number of heat source targets, the operating status data, and the passenger distribution coefficient of the vehicle.

[0029] S2. Introduce a non-negative matrix Z to represent the missing value data structure, adjust the objective function of the NMF model according to the non-negative matrix Z, and determine the corresponding iterative optimization strategy;

[0030] In the nonnegative matrix Z, 1 represents missing values ​​and 0 represents non-missing values;

[0031] S3. Solve the NMF model based on the iterative optimization strategy. In each iteration, fix other matrices and update one matrix until the iteration termination condition is met, and obtain the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z.

[0032] Wherein, the non-negative matrix W is the basis image matrix, which contains features or basis extracted from the non-negative matrix V, and each column of the matrix is ​​regarded as a basis feature or basis vector;

[0033] The non-negative matrix H is a coefficient matrix, which contains coefficients for linear combination of basis vectors in the non-negative matrix W, each row of the matrix corresponds to a data point in the non-negative matrix V, and each column of the matrix corresponds to a basis vector in the non-negative matrix W;

[0034] S4, filling in missing values in the preprocessed passenger distribution coefficients and vehicle data according to the decomposed non-negative matrix W, the non-negative matrix H and the non-negative matrix Z, and calculating filling data corresponding to the missing values;

[0035] S5, smoothing and verifying the calculated filling data to ensure the accuracy and reasonableness of the filling data.

[0036] Preferably, a non-negative matrix Z representing the data structure of the missing values is introduced in S2, the objective function of the NMF model is adjusted according to the non-negative matrix Z, and a corresponding iterative optimization strategy is determined, including:

[0037] S21, determining the data structure of the missing values, and introducing a non-negative matrix Z representing the data structure of the missing values;

[0038] S22, adjusting the objective function of the NMF model according to the non-negative matrix Z:

[0039]

[0040] Wherein, V, W, H and Z are the non-negative matrix V, the non-negative matrix W, the non-negative matrix H and the non-negative matrix Z respectively, V nmv , H nmv are the observed non-missing value parts in the non-negative matrix V and the non-negative matrix H respectively, λ is a regularization parameter, represents Hadamard product, represents the square of Frobenius norm;

[0041] S23, taking the alternating minimization strategy as the iterative optimization strategy for solving the NMF model.

[0042] Preferably, the NMF model is solved based on the iterative optimization strategy in S3, in each iteration process, one matrix is updated while the other matrices are fixed, until the iteration termination condition is met, and the decomposed non-negative matrix W, the non-negative matrix H and the non-negative matrix Z are obtained, including:

[0043] S31, solving the NMF model based on the alternating minimization strategy;

[0044] S32, in each iteration process, updating according to the updating order of the non-negative matrix W, the non-negative matrix H, the non-negative matrix Z and the filling data;

[0045] S33, judging whether an iteration termination condition is met, if the iteration termination condition is not met, returning to S31, otherwise taking the current non-negative matrix W, non-negative matrix H and non-negative matrix Z as the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z;

[0046] The iteration termination condition includes that a target function value converges to below a certain threshold value, or the number of iterations reaches a preset maximum number.

[0047] Preferably, the control terminal utilizes the control decision module to evaluate the passenger flow and the vehicle operation condition according to the processed received data, and formulates a control scheduling strategy, and sends the control scheduling strategy to the vehicle terminal through the ground base station, including:

[0048] The control terminal utilizes the control decision module to evaluate the passenger flow according to the processed passenger distribution coefficient and the heat source area and the number of heat source targets of each car collected by the infrared sensor in the vehicle data;

[0049] The control terminal utilizes the control decision module to input the running state data of each car collected by other sensors in the processed vehicle data and the corresponding position information when the running state data is acquired by other sensors into a pre-trained vehicle operation condition judgment model, and evaluates the vehicle operation condition;

[0050] The control decision module comprehensively analyzes the passenger flow evaluation result and the vehicle operation condition evaluation result, and formulates a control scheduling strategy, and the control terminal sends the control scheduling strategy to the vehicle terminal through the ground base station.

[0051] Preferably, before the control terminal utilizes the control decision module to input the running state data of each car collected by other sensors in the processed vehicle data and the corresponding position information when the running state data is acquired by other sensors into a pre-trained vehicle operation condition judgment model, and evaluates the vehicle operation condition, including:

[0052] S1, dividing a historical data set into a training set, a validation set and a test set according to a preset proportion;

[0053] S2, setting a loss function and an optimizer of the vehicle operation condition judgment model;

[0054] S3, inputting the training set into the vehicle operation condition judgment model for model training;

[0055] S4, calculating a loss value based on the loss function, and updating model parameters according to the loss value and network gradient information by the optimizer;

[0056] S5, if the loss value is less than the preset threshold value, the model training is ended, the current vehicle operation condition judgment model is the trained vehicle operation condition judgment model, otherwise, returning to S3, continuing to train the model by using the training set;

[0057] S6, input the verification set into the trained vehicle operation condition judgment model, evaluate the generalization ability of the model by observing the performance on the verification set, and optimize the hyperparameters and structure of the model;

[0058] S7, input the test set into the optimized vehicle operation condition judgment model, and evaluate the performance of the model.

[0059] (Three) beneficial effects

[0060] Compared with the prior art, the intelligent subway monitoring system provided by the application has the following beneficial effects:

[0061] 1) The control terminal receives the data transmitted by the ground base station, first pre-processes the received data through the data processing module, and then fills in the missing values in the passenger distribution coefficient and vehicle data after preprocessing by the data processing module using the improved non-negative matrix factorization NMF. In the improved non-negative matrix factorization NMF, a matrix representing the data structure of the missing values is introduced to improve the accuracy of filling in the data, while maintaining the consistency of the filled data with the original data, thereby accurately filling in the missing values in the time-based vehicle data, providing data support for the control decision module to develop control scheduling strategies.

[0062] 2) The sensor module collects vehicle data, the positioning module collects the position information corresponding to the vehicle data collected by the sensor module, the passenger distribution coefficient calculation module calculates the passenger distribution coefficient of the vehicle according to the vehicle data, and the vehicle terminal encapsulates the passenger distribution coefficient, vehicle data and corresponding position information and sends it to the ground base station. After the data processing module processes the data received by the control terminal from the ground base station, accurate and complete received data can be obtained, and the control decision module can accurately evaluate the passenger flow according to the processed passenger distribution coefficient and the infrared sensor collected area of the heat source region and the number of heat source targets in each car in the vehicle data.

[0063] 3) At the same time, the control decision module inputs the processed vehicle data and the corresponding position information of the running state data collected by other sensors in each car into the pre-trained vehicle operation condition judgment model, which can accurately evaluate the running condition of the vehicle. The control decision module comprehensively analyzes the passenger flow evaluation result and the vehicle running condition evaluation result, and develops a control scheduling strategy to ensure the accuracy and rationality of the control scheduling strategy, thereby accurately and efficiently controlling and scheduling the subway vehicle depot. BRIEF DESCRIPTION OF DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0065] Figure 1 The system schematic diagram of the present application;

[0066] Figure 2 The flowchart schematic diagram of the present application. DETAILED DESCRIPTION

[0067] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0068] A smart subway monitoring system, as shown in Figure 1 and Figure 2 , comprises a vehicle terminal and a control terminal;

[0069] The vehicle terminal collects vehicle data through a sensor module, collects corresponding position information of the vehicle data obtained by the sensor module through a positioning module, calculates the passenger distribution coefficient of the vehicle according to the vehicle data by using a passenger distribution coefficient calculation module, and sends the passenger distribution coefficient, the vehicle data and the corresponding position information to a ground base station after encapsulation.

[0070] The control terminal receives the data transmitted by the ground base station, processes the received data through a data processing module, evaluates the passenger flow and vehicle operation according to the processed received data by using a control decision module, and formulates a control scheduling strategy, which is sent to the vehicle terminal through the ground base station.

[0071] Among them, the data processing module uses an improved non-negative matrix factorization NMF to fill in the missing values in the passenger distribution coefficient and the vehicle data. In the improved non-negative matrix factorization NMF, a matrix representing the data structure of the missing values is introduced to improve the accuracy of the filled data, while maintaining the consistency of the filled data and the original data.

[0072] The technical scheme has the advantages that the control terminal receives data transmitted by the ground base station, and first pre-processes the received data through the data processing module, and then fills in missing values in the passenger distribution coefficient and the vehicle data after pre-processing by the data processing module using an improved non-negative matrix factorization (NMF), wherein a matrix representing the data structure of the missing values is introduced to improve the accuracy of filling in the data while maintaining the consistency of the filled data with the original data, thereby accurately filling in the missing values in the time-series-based vehicle data and providing data support for the control decision module to formulate a control scheduling strategy.

[0073] In the technical scheme of the present application, for the vehicle end:

[0074] ① The vehicle terminal collects vehicle data through the sensor module, and collects corresponding position information of the vehicle data collected by the sensor module through the positioning module, including:

[0075] The vehicle terminal collects vehicle data through the sensor module installed in each carriage, and the vehicle data includes the area of the heat source region of each carriage collected by the infrared sensor, the number of heat source targets, and the running state data of each carriage collected by other sensors;

[0076] Meanwhile, the vehicle terminal collects corresponding position information of the vehicle data collected by the sensor module through the positioning module installed in each carriage, and the position information includes the carriage number, the nearest station information and the GPS position information.

[0077] ② The vehicle terminal calculates the passenger distribution coefficient of the vehicle according to the vehicle data by using the passenger distribution coefficient calculation module, including:

[0078] The vehicle terminal calculates the passenger distribution coefficient of the vehicle according to the area of the heat source region and the number of heat source targets of each carriage by using the passenger distribution coefficient calculation module, using the following formula:

[0079]

[0080] wherein PDC i is the passenger distribution coefficient of the vehicle i, S ij and S' ij are the area of the heat source region and the maximum area of the passengers in carriage j of the vehicle i, N ij and N' ij are the number of heat source targets and the maximum number of passengers in carriage j of the vehicle i, and ω j is the weight coefficient of carriage j of the vehicle i, which is set according to the functional attributes of the surrounding building facilities of the vehicle arriving at the station, j∈[1,n], and n is the number of carriages of the vehicle i.

[0081] The vehicle terminal encapsulates the passenger distribution coefficient, vehicle data and corresponding position information and sends them to the ground base station, including:

[0082] The vehicle terminal encapsulates the passenger distribution coefficient, vehicle data and corresponding position information, and sends them to the ground base station through a wireless communication module.

[0083] The ground base station receives the encapsulated data, decodes and checks them to ensure their accuracy and integrity, and transmits them to the control terminal.

[0084] In the technical solution of the present application, the control center:

[0085] The control terminal receives the data transmitted by the ground base station, processes the received data through a data processing module, including:

[0086] The control terminal receives the data transmitted by the ground base station, and pre-processes the received data through a data processing module.

[0087] The data processing module uses an improved non-negative matrix factorization (NMF) to fill in the missing values in the pre-processed passenger distribution coefficient and vehicle data. In the improved NMF, a matrix representing the data structure of missing values is introduced to improve the accuracy of the filled data while maintaining the consistency of the filled data with the original data.

[0088] The data processing module pre-processes the received data, including removing outliers and smoothing.

[0089] Specifically, the data processing module uses an improved non-negative matrix factorization (NMF) to fill in the missing values in the pre-processed passenger distribution coefficient and vehicle data, including:

[0090] S1, arrange the pre-processed passenger distribution coefficient and vehicle data into a non-negative matrix V.

[0091] In the non-negative matrix V, the missing values are represented by 0. Each row of the matrix represents a time point, and each column of the matrix represents a data category, including the heat source area of each car, the number of heat source targets, the running state data, and the passenger distribution coefficient of the vehicle.

[0092] S2, introduce a non-negative matrix Z representing the data structure of missing values, adjust the objective function of the NMF model according to the non-negative matrix Z, and determine the corresponding iterative optimization strategy.

[0093] In the non-negative matrix Z, 1 represents a missing value and 0 represents a non-missing value.

[0094] S3, solving the NMF model based on an iterative optimization strategy, in each iteration process, fixing other matrices, updating one matrix, until the iteration termination condition is met, obtaining the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z;

[0095] wherein the non-negative matrix W is a basic image matrix, which contains the features or bases extracted from the non-negative matrix V, and each column of the matrix is regarded as a basic feature or base vector;

[0096] The non-negative matrix H is a coefficient matrix, which contains the coefficients for linearly combining the base vectors in the non-negative matrix W, each row of the matrix corresponds to a data point in the non-negative matrix V, and each column of the matrix corresponds to a base vector in the non-negative matrix W;

[0097] S4, filling the missing values in the preprocessed passenger distribution coefficients and vehicle data according to the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z, and calculating the filling data corresponding to the missing values;

[0098] S5, smoothing and verifying the calculated filling data to ensure the accuracy and reasonableness of the filling data.

[0099] Specifically, a non-negative matrix Z representing the data structure of the missing values is introduced in S2, the objective function of the NMF model is adjusted according to the non-negative matrix Z, and the corresponding iterative optimization strategy is determined, including:

[0100] S21, determining the data structure of the missing values, and introducing a non-negative matrix Z representing the data structure of the missing values;

[0101] S22, adjusting the objective function of the NMF model according to the non-negative matrix Z:

[0102]

[0103] wherein V, W, H, Z are non-negative matrix V, non-negative matrix W, non-negative matrix H, non-negative matrix Z, V nmv , H nmv are the observed non-missing value parts in the non-negative matrix V and the non-negative matrix H, λ is a regularization parameter, represents Hadamard product, represents the square of Frobenius norm;

[0104] S23, taking the alternating minimization strategy as the iterative optimization strategy for solving the NMF model.

[0105] Specifically, the NMF model is solved based on an iterative optimization strategy in S3, in each iteration process, one matrix is updated while fixing other matrices, until the iteration termination condition is met, to obtain the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z, including:

[0106] S31, the NMF model is solved based on an alternating minimization strategy;

[0107] S32, in each iteration process, the update is performed according to the update order of the non-negative matrix W, the non-negative matrix H, the non-negative matrix Z and the filled data;

[0108] S33, it is judged whether the iteration termination condition is met, if the iteration termination condition is not met, S31 is returned, otherwise the current non-negative matrix W, non-negative matrix H and non-negative matrix Z are taken as the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z;

[0109] Wherein, the iteration termination condition includes that the objective function value converges to below a certain threshold, or the number of iterations reaches a preset maximum number.

[0110] ② The control terminal utilizes the control decision module to evaluate the passenger flow and the vehicle operation condition according to the processed received data, and formulates a control scheduling strategy, which is sent to the vehicle terminal through the ground base station, including:

[0111] The control terminal utilizes the control decision module to evaluate the passenger flow according to the processed passenger distribution coefficient, and the area of the heat source region and the number of heat source targets of each car collected by the infrared sensor in the vehicle data;

[0112] The control terminal utilizes the control decision module to input the running state data of each car collected by other sensors in the processed vehicle data, and the position information corresponding to the acquisition of the running state data by other sensors into the pre-trained vehicle operation condition judgment model, to evaluate the vehicle operation condition;

[0113] The control decision module comprehensively analyzes the passenger flow evaluation result and the vehicle operation condition evaluation result, and formulates a control scheduling strategy, which is sent to the vehicle terminal by the control terminal through the ground base station.

[0114] Specifically, before the control terminal utilizes the control decision module to input the running state data of each car collected by other sensors in the processed vehicle data, and the position information corresponding to the acquisition of the running state data by other sensors into the pre-trained vehicle operation condition judgment model to evaluate the vehicle operation condition, including:

[0115] S1, divide the historical data set into a training set, a validation set and a test set according to a preset proportion;

[0116] S2, set a loss function and an optimizer of a vehicle operation condition judgment model;

[0117] S3, input the training set into the vehicle operation condition judgment model for model training;

[0118] S4, calculate a loss value based on the loss function, and update model parameters according to the loss value and network gradient information by the optimizer;

[0119] S5, if the loss value is less than a preset threshold, the model training is ended, and the current vehicle operation condition judgment model is the trained vehicle operation condition judgment model, otherwise, return to S3 to continue model training by using the training set;

[0120] S6, input the verification set into the trained vehicle operation condition judgment model, evaluate the generalization ability of the model by observing the performance on the verification set, and optimize the hyperparameters and structure of the model;

[0121] S7, input the test set into the optimized vehicle operation condition judgment model, and evaluate the performance of the model.

[0122] The above technical scheme, the sensor module collects vehicle data, the positioning module collects position information corresponding to the vehicle data collected by the sensor module, the passenger distribution coefficient calculation module calculates the passenger distribution coefficient of the vehicle according to the vehicle data, the vehicle terminal encapsulates and sends the passenger distribution coefficient, the vehicle data and the corresponding position information to the ground base station, the control terminal receives the data transmitted by the ground base station, and after data processing by the data processing module, accurate and complete received data can be obtained, and the control decision module can accurately evaluate the passenger flow according to the processed passenger distribution coefficient and the infrared sensor area of the heat source region in the vehicle data, the number of heat source targets in each car, and the number of heat source targets in each car.

[0123] At the same time, the control decision module inputs the running state data of each car collected by other sensors in the processed vehicle data and the position information corresponding to the running state data collected by other sensors into the pre-trained vehicle operation condition judgment model, which can accurately evaluate the running state of the vehicle, the control decision module comprehensively analyzes the passenger flow evaluation result and the vehicle operation condition evaluation result, and formulates a control scheduling strategy to ensure the accuracy and rationality of the control scheduling strategy, thereby accurately and efficiently controlling and scheduling the metro vehicle depot.

[0124] The above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can be modified, or some technical features can be replaced by equivalent features; and these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A smart subway monitoring system, characterized in that: The vehicle terminal and the control terminal are included. The vehicle terminal collects vehicle data through a sensor module and collects position information corresponding to the vehicle data collected by the sensor module through a positioning module, calculates a passenger distribution coefficient of the vehicle according to the vehicle data by using a passenger distribution coefficient calculation module, and encapsulates the passenger distribution coefficient, the vehicle data and the corresponding position information and then sends them to a ground base station. The control terminal receives data transmitted by the ground base station, processes the received data by using a data processing module, evaluates passenger flow and vehicle operation according to the processed received data by using a control decision module, and formulates a control scheduling strategy, which is sent to the vehicle terminal through the ground base station. The control terminal receives data transmitted by the ground base station, processes the received data by using a data processing module, and includes the following steps. The control terminal receives data transmitted by the ground base station, processes the received data by using a data processing module, and includes the following steps. The data processing module uses an improved non-negative matrix factorization (NMF) to fill in missing values in the preprocessed passenger distribution coefficient and vehicle data, and the improved NMF introduces a matrix representing the data structure of the missing values to improve the accuracy of the filled data while maintaining the consistency of the filled data with the original data. The data processing module uses an improved non-negative matrix factorization (NMF) to fill in missing values in the preprocessed passenger distribution coefficient and vehicle data, and the improved NMF introduces a matrix representing the data structure of the missing values to improve the accuracy of the filled data while maintaining the consistency of the filled data with the original data. The data processing module uses an improved non-negative matrix factorization (NMF) to fill in missing values in the preprocessed passenger distribution coefficient and vehicle data, and the improved NMF introduces a matrix representing the data structure of the missing values to improve the accuracy of the filled data while maintaining the consistency of the filled data with the original data. S1, the preprocessed passenger distribution coefficient and vehicle data are arranged into a non-negative matrix V. The missing values in the non-negative matrix V are represented by 0, each row of the matrix represents a time point, and each column of the matrix represents a data category, including the hot source area of each car, the number of hot source targets, the running state data, and the passenger distribution coefficient of the vehicle. S2, a non-negative matrix Z representing the data structure of the missing values is introduced, the objective function of the NMF model is adjusted according to the non-negative matrix Z, and the corresponding iterative optimization strategy is determined. In the non-negative matrix Z, 1 represents a missing value and 0 represents a non-missing value. S3, based on the iterative optimization strategy, the NMF model is solved, in each iteration process, other matrices are fixed, and one matrix is updated until the iteration termination condition is met, and the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z are obtained. The non-negative matrix W is a basic image matrix, which contains features or bases extracted from the non-negative matrix V, and each column of the matrix is regarded as a basic feature or basis vector. The non-negative matrix H is a coefficient matrix, which contains coefficients for linear combination of basis vectors in the non-negative matrix W, each row of the matrix corresponds to a data point in the non-negative matrix V, and each column of the matrix corresponds to a basis vector in the non-negative matrix W. S4, according to the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z, the missing values in the preprocessed passenger distribution coefficient and vehicle data are filled in, and the filled data corresponding to the missing values is calculated. S5, smoothing and verifying the calculated imputed data to ensure the accuracy and reasonableness of the imputed data; In S2, a non-negative matrix Z representing the data structure of missing values is introduced, the objective function of the NMF model is adjusted according to the non-negative matrix Z, and the corresponding iterative optimization strategy is determined, including: S21, determining the data structure of missing values, and introducing a non-negative matrix Z representing the data structure of missing values; S22, adjusting the objective function of the NMF model according to the non-negative matrix Z: where V, W, H, Z are non-negative matrices, V nmv , H nmv are the observed non-missing parts of non-negative matrices V, H, respectively, λ is a regularization parameter, denotes the Hadamard product, denotes the square of the Frobenius norm; S23, taking the alternating minimization strategy as the iterative optimization strategy for solving the NMF model.

2. The intelligent subway monitoring system according to claim 1, wherein: The vehicle terminal collects vehicle data through the sensor module, and simultaneously collects corresponding position information when the sensor module collects vehicle data through the positioning module, including: The vehicle terminal collects vehicle data through the sensor module installed in each car compartment, and the vehicle data includes the area of the heat source region and the number of heat source targets collected by the infrared sensor in each car compartment, as well as the running state data collected by other sensors in each car compartment. At the same time, the vehicle terminal collects the position information corresponding to the vehicle data collected by the sensor module through the positioning module installed in each car compartment, and the position information includes the car compartment number, the nearest station information and the GPS position information. 3.The smart subway monitoring system according to claim 2, characterized in that: The vehicle terminal calculates the passenger distribution coefficient of the vehicle according to the vehicle data by using the passenger distribution coefficient calculation module, including: The vehicle terminal calculates the passenger distribution coefficient of the vehicle according to the area of the heat source region and the number of heat source targets in each car compartment by using the passenger distribution coefficient calculation module, using the following formula: Wherein, PDC i is the passenger distribution coefficient of vehicle i, S ij , S' ij are the heat source area and the maximum passenger accommodation area of the car j in the vehicle i respectively, N ij , N' ij are the heat source target number and the maximum passenger accommodation number of the car j in the vehicle i respectively, ω j is the weight coefficient of the car j in the vehicle i, which is set according to the functional attribute of the building facilities around the station reached by the vehicle, j∈[1,n], and n is the number of cars of the vehicle i.

4. The intelligent subway monitoring system according to claim 3, characterized in that: The vehicle terminal encapsulates the passenger distribution coefficient, vehicle data and corresponding position information and sends them to the ground base station, including: After encapsulating the passenger distribution coefficient, vehicle data and corresponding position information, the vehicle terminal sends them to the ground base station through the wireless communication module. After receiving the encapsulated data, the ground base station decodes and verifies the data to ensure the accuracy and integrity of the data, and transmits the data to the control terminal.

5. The intelligent subway monitoring system according to claim 1, wherein: In S3, the NMF model is solved based on the iterative optimization strategy. In each iteration process, one matrix is updated while the other matrices are fixed, until the iteration termination condition is met, and the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z are obtained, including: S31, solving the NMF model based on the alternating minimization strategy; S32, updating in each iteration process according to the update order of the non-negative matrix W, non-negative matrix H, non-negative matrix Z and imputed data; S33, determining whether the iteration termination condition is met, if not, returning to S31, otherwise, taking the current non-negative matrix W, non-negative matrix H and non-negative matrix Z as the decomposed non-negative matrix W, non-negative matrix H and non-negative matrix Z; The iteration termination condition includes that the objective function value converges to a certain threshold value or the iteration number reaches a preset maximum number. 6.The smart subway monitoring system according to claim 1, characterized in that: The control terminal evaluates the passenger flow and vehicle operation condition according to the processed received data by using the control decision module, and formulates a control scheduling strategy, which is sent to the vehicle terminal through the ground base station, including: The control terminal utilizes the control decision module to evaluate the passenger flow according to the processed passenger distribution coefficient and the heat source area and the number of heat source targets of each car collected by the infrared sensor in the vehicle data; The control terminal utilizes the control decision module to evaluate the vehicle operation condition by inputting the running state data of each car collected by other sensors in the processed vehicle data and the corresponding position information obtained by other sensors into the pre-trained vehicle operation condition judgment model; The control decision module comprehensively analyzes the passenger flow evaluation result and the vehicle operation condition evaluation result, and formulates a control scheduling strategy, and the control terminal sends the control scheduling strategy to the vehicle terminal through the ground base station. 7.The smart subway monitoring system according to claim 6, characterized in that: Before the control terminal utilizes the control decision module to evaluate the vehicle operation condition by inputting the running state data of each car collected by other sensors in the processed vehicle data and the corresponding position information obtained by other sensors into the pre-trained vehicle operation condition judgment model, the method comprises the following steps: S1, divide the historical data set into a training set, a validation set and a test set according to a preset proportion; S2, set the loss function and the optimizer of the vehicle operation condition judgment model; S3, input the training set into the vehicle operation condition judgment model for model training; S4, calculate the loss value based on the loss function, and update the model parameters according to the loss value and the network gradient information by the optimizer; S5, if the loss value is less than a preset threshold, the model training is ended, and the current vehicle operation condition judgment model is the trained vehicle operation condition judgment model, otherwise, return to S3 to continue model training by using the training set; S6, input the validation set into the trained vehicle operation condition judgment model, evaluate the generalization ability of the model by observing the performance on the validation set, and optimize the hyperparameters and structure of the model; S7, input the test set into the optimized vehicle operation condition judgment model, and evaluate the performance of the model.

Images