Train valve control method, device, equipment, medium, program product and train

Abstract

The invention provides a train valve control method which can be applied to the technical field of rail transit and vehicle pressure wave protection. The method comprises the steps that vehicle-mounted information and a mapping relation corresponding to a train compartment are obtained, the vehicle-mounted information comprises operation information, pressure information and air quality information, and the mapping relation represents the corresponding relation between valve control information of the train compartment and the vehicle-mounted information; the operation information, the pressure information and the air quality information are input into a prediction model, prediction results are output, and the prediction results comprise a pressure prediction result and an air quality prediction result; and switching or keeping the state of the valve based on the prediction result and the mapping relation. The invention further provides a train valve control device, equipment, a medium, a program product and a train.

Description

Technical Field

[0001] The present disclosure relates to the technical field of rail transit and vehicle pressure wave protection, and in particular to a train valve control method, device, equipment, medium, program product and train. Background Art

[0002] When high-speed trains run on the line, it is inevitable that they will pass through tunnels, and there may even be situations where two trains intersect in the tunnel. Due to the limitations of tunnel space and the compressibility of air, when a train enters a tunnel, it will produce more violent compression waves and expansion waves, of which the compression wave is positive pressure and the expansion wave is negative pressure. These two types of pressure waves propagate at the speed of sound in the tunnel and are reflected back to the inside of the tunnel with some capacity at the tunnel exit and continue to act on the car body, causing violent pressure fluctuations on the surface of the car body. Therefore, reasonable pressure wave control logic is crucial to the impact of train pressure fluctuations. In the related art, the train pressure wave protection control method mainly includes installing a pressurization protection device at the air outlet position to compress the external air; or by entering all route information in advance, according to the positioning technology, when it is judged that the train needs to enter the tunnel, the pressure wave protection system is turned on.

[0003] The inventors discovered that in the related art, the control method for pressure waves has the following defects: the traditional pressure wave control logic is not suitable for the current speed and line conditions, especially when there are complex tunnel groups in mountainous areas, and the quality of the air in the car during the whole process is ignored, or the pressure fluctuations are too large, affecting the comfort and riding experience of the passengers. Summary of the invention

[0004] In view of the above problems, the present disclosure provides a train valve control method, device, equipment, storage medium, program product and train.

[0005] According to the first aspect of the present disclosure, a train valve control method is provided, comprising: obtaining on-board information and a mapping relationship corresponding to a train compartment, wherein the on-board information comprises operating information, pressure information and air quality information, and the mapping relationship represents the correspondence between the valve control information of the train compartment and the on-board information; inputting the operating information, the pressure information and the air quality information into a prediction model, outputting a prediction result, wherein the prediction result comprises a pressure prediction result and an air quality prediction result; and switching or maintaining the state of the valve based on the prediction result and the mapping relationship.

[0006] According to an embodiment of the present disclosure, the above method also includes: obtaining feedback information corresponding to the above state, wherein the above feedback information represents the difference result between the above state and a preset state threshold; and in the case where the above difference result represents that the above state is less than the above preset state threshold, using a preset update strategy to update the above prediction model to obtain a target prediction model.

[0007] According to an embodiment of the present disclosure, the above-mentioned prediction model includes an initial interaction layer; updating the above-mentioned prediction model using a preset update strategy to obtain a target prediction model includes: updating the above-mentioned initial interaction layer using the above-mentioned preset update strategy to obtain a target interaction layer, wherein the above-mentioned target interaction layer includes a first interaction layer and a second interaction layer, and the above-mentioned second interaction layer includes multiple feature vector processing sublayers; and updating the above-mentioned prediction model based on the above-mentioned second interaction layer to obtain the above-mentioned target prediction model.

[0008] According to an embodiment of the present disclosure, the above method also includes: classifying the above operation information, the above pressure information and the above air quality information using preset classification rules to obtain multiple feature domain information; processing the multiple feature domain information to obtain processing results; and determining the target prediction result based on the above mapping relationship and the above processing results, the above target prediction result characterizing the state of the above valve in the target time period.

[0009] According to an embodiment of the present disclosure, multiple feature domain information are processed to obtain processing results, including: mapping the multiple feature domain information to a low-dimensional continuous space to obtain multiple feature vectors; determining the interactive feature information between the multiple feature vectors, wherein the interactive feature information includes the first interactive information obtained using the first interactive layer and the second interactive information obtained using the second interactive layer; and updating the first interactive information and the second interactive information based on a preset conversion rule to obtain the processing results.

[0010] According to an embodiment of the present disclosure, switching or maintaining the state of the valve based on the above-mentioned prediction results and the above-mentioned mapping relationship includes: in a case where the above-mentioned prediction results and the above-mentioned mapping relationship indicate that the state of the above-mentioned valve is switched during the target time period, switching the state of the above-mentioned valve based on a preset switching rule, wherein the above-mentioned preset switching rule includes at least one of a car-by-car switching rule and a time-by-time switching rule.

[0011] According to an embodiment of the present disclosure, the above-mentioned pressure information includes pressure difference information, the above-mentioned operation information includes speed information, the above-mentioned mapping relationship includes a first mapping relationship and a second mapping relationship, the above-mentioned first mapping relationship represents the correspondence between the above-mentioned pressure difference information and the above-mentioned valve control information, and the above-mentioned second mapping relationship represents the correspondence between the above-mentioned speed information and the above-mentioned valve control information.

[0012] According to an embodiment of the present disclosure, the above-mentioned prediction model is obtained by training through the following operations: obtaining sample initial on-board information corresponding to a sample train, wherein the above-mentioned sample initial on-board information includes initial operation information, initial pressure information and initial air quality information; processing the above-mentioned sample initial on-board information to obtain sample on-board information, wherein the amount of information of the above-mentioned sample on-board information is greater than the amount of information of the above-mentioned sample initial on-board information; and using the above-mentioned sample on-board information to train an initial prediction model to obtain the above-mentioned prediction model, wherein the above-mentioned sample on-board information includes a label corresponding to the above-mentioned sample on-board information.

[0013] According to an embodiment of the present disclosure, the above-mentioned sample vehicle information is used to train an initial prediction model to obtain the above-mentioned prediction model, including: using the above-mentioned initial prediction model to classify the above-mentioned sample vehicle information to obtain multiple sample feature domain information; processing the multiple sample feature domain information to obtain sample processing results; determining sample prediction results based on the above-mentioned sample processing results and sample mapping relationships; determining the difference between the above-mentioned sample prediction results and the verification results; and when the above-mentioned difference meets a preset threshold, determining the above-mentioned initial prediction model as the above-mentioned prediction model.

[0014] According to an embodiment of the present disclosure, multiple sample feature domain information are processed to obtain sample processing results, including: mapping multiple sample feature domain information to a low-dimensional continuous space to obtain multiple sample feature information; determining sample interaction feature information between multiple sample feature information; and performing nonlinear transformation on the sample interaction feature information to obtain the sample processing results.

[0015] According to an embodiment of the present disclosure, the above-mentioned sample initial vehicle-mounted information is processed to obtain sample vehicle-mounted information, including: using a preset processing strategy to process the initial abnormal information in the above-mentioned sample initial vehicle-mounted information to obtain abnormal information; and updating the above-mentioned sample initial vehicle-mounted information based on the above-mentioned abnormal information to obtain the above-mentioned sample vehicle-mounted information.

[0016] According to an embodiment of the present disclosure, the preset processing strategy includes a random noise strategy; the initial abnormal information in the above-mentioned sample initial vehicle-mounted information is processed using the preset processing strategy to obtain abnormal information, including: determining the noise type and noise parameters corresponding to the above-mentioned initial abnormal information; generating noise information based on the above-mentioned noise type and the above-mentioned noise parameters; and generating the above-mentioned abnormal information using the above-mentioned noise information and the initial abnormal information.

[0017] According to an embodiment of the present disclosure, the preset processing strategy also includes a translation strategy; using the preset processing strategy to process the initial abnormal information in the above-mentioned sample initial vehicle-mounted information to obtain the abnormal information, and also includes: determining a translation value corresponding to the above-mentioned initial abnormal information, wherein the above-mentioned translation value represents the movement amplitude of the above-mentioned initial abnormal information over the target historical time period; and translating the above-mentioned initial abnormal information based on the above-mentioned movement amplitude to obtain the above-mentioned abnormal information.

[0018] The second aspect of the present disclosure provides a train valve control device, including: an information acquisition module, used to acquire on-board information and a mapping relationship corresponding to a train car, wherein the on-board information includes operating information, pressure information and air quality information, and the mapping relationship represents the correspondence between the valve control information of the train car and the on-board information; a prediction result output module, used to input the operating information, the pressure information and the air quality information into a prediction model, and output a prediction result, wherein the prediction result includes a pressure prediction result and an air quality prediction result; and a switching or maintaining module, used to switch or maintain the state of the valve based on the prediction result and the mapping relationship.

[0019] The third aspect of the present disclosure provides a train, comprising: a train compartment; a monitoring device for monitoring the on-board information of the train compartment; the valve in the above method, for controlling the exchange between the air inside the train compartment and the outside air; and the above train valve control device.

[0020] A fourth aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above method.

[0021] The fifth aspect of the present disclosure also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the above method.

[0022] The sixth aspect of the present disclosure also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0023] According to the train valve control method, device, equipment, medium, program product and train provided by the present invention, by inputting the real-time operation information, pressure information and air quality information into the prediction model, the pressure prediction result and air quality prediction result inside the car are obtained in real time, so that the valve state can be flexibly switched or maintained based on the mapping relationship between the valve control information and the on-board information. Since the prediction result is generated in real time based on the prediction model for different operation information, pressure information and air quality information, it satisfies the control of pressure fluctuations under different speed levels and different line conditions (such as complex tunnel groups in mountainous areas), avoids the situation where the passenger comfort is affected by excessive pressure fluctuations in the car or poor air quality, and further improves the passenger riding experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0025] Figure 1 The application scenario diagram of the train valve control method, device, equipment, storage medium, and program product according to the embodiments of the present disclosure is schematically shown;

[0026] Figure 2 A time-varying curve diagram of pressure fluctuation on the outer surface of a train body passing through a tunnel according to an embodiment of the present disclosure is schematically shown;

[0027] Figure 3 A flow chart of a train valve control method according to an embodiment of the present disclosure is schematically shown;

[0028] Figure 4 The schematic diagram of the structure of the target prediction model according to the embodiment of the present disclosure is shown;

[0029] Figure 5 A diagram schematically shows a train in-car pressure wave protection system according to an embodiment of the present disclosure;

[0030] Figure 6 The schematic diagram of the structure of the initial prediction model according to the embodiment of the present disclosure is shown;

[0031] Figure 7 The structure block diagram of the train valve control device according to the embodiment of the present disclosure is schematically shown;

[0032] Figure 8 A block diagram of an electronic device suitable for implementing a train valve control method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0033] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0034] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0035] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0036] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0037] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0038] In the process of conceiving the present invention, the inventors discovered that in the related art, regarding the control method of pressure waves inside a train, the traditional pressure wave control logic is not suitable for the current speed and line conditions, especially when there are complex tunnel groups in mountainous areas. It ignores the air quality inside the car during the entire process, or the pressure fluctuations are too large, affecting the comfort and riding experience of the passengers.

[0039] In view of this, the present invention obtains the pressure prediction results and air quality prediction results inside the car in real time by inputting the real-time operation information, pressure information and air quality information into the prediction model, so as to flexibly switch or maintain the valve state based on the mapping relationship between the valve control information and the vehicle-mounted information. Since the prediction result is generated in real time based on the prediction model for different operation information, pressure information and air quality information, it satisfies the control of pressure fluctuations in the car under different speed levels and different line conditions (such as mountainous areas and complex tunnel groups), avoids the situation where the passenger comfort is affected by excessive pressure fluctuations in the car or poor air quality, and further improves the passenger riding experience.

[0040] Embodiments of the present disclosure provide a train valve control method, apparatus, device, storage medium, program product and train, the method comprising obtaining on-board information and a mapping relationship corresponding to a train car, wherein the on-board information comprises operating information, pressure information and air quality information, and the mapping relationship represents the correspondence between the valve control information of the train car and the on-board information; inputting the operating information, pressure information and air quality information into a prediction model, outputting prediction results, wherein the prediction results comprise pressure prediction results and air quality prediction results; and switching or maintaining the state of the valve based on the prediction results and the mapping relationship.

[0041] Figure 1 The application scenario diagram of the train valve control method, device, equipment, storage medium, and program product according to the embodiments of the present disclosure is schematically shown.

[0042] like Figure 1 As shown, the application scenario according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, a server 105, and a monitoring device 106. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, the server 105, and the monitoring device 106. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0043] The monitoring device 106 may be a device installed on the train for real-time monitoring of train onboard information, including in-car and out-car pressure monitoring devices, in-car air quality monitoring devices, speed monitoring devices, and line information monitoring devices, such as pressure sensors, carbon dioxide concentration sensors, etc. It is understandable that the monitoring device 106 may be connected to the server 105 or a server cluster via the network 104 to achieve real-time data processing.

[0044] The user may use at least one of the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0045] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0046] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0047] It should be noted that the train valve control method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the train valve control device provided in the embodiment of the present disclosure can generally be arranged in the server 105. The train valve control method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the train valve control device provided in the embodiment of the present disclosure can also be arranged in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0048] It should be understood that Figure 1 The number of terminal devices, networks, monitoring devices and servers in the embodiment is only for illustration. Any number of terminal devices, networks, monitoring devices and servers may be provided according to implementation requirements.

[0049] It is understandable that the air conditioning system is an important device to ensure the comfortable temperature and air quality of passengers in the carriage. In addition to controlling temperature and humidity, it also has ventilation and air purification functions. The air conditioning system will introduce an appropriate amount of fresh air during the operation of the train to ensure fresh air in the car and prevent the air from becoming polluted due to long-term sealing. The exchange of fresh air with the outside world is achieved through the fresh air valve and exhaust valve on the car body.

[0050] Figure 2 A time variation curve of pressure fluctuation on the outer surface of a train body passing through a tunnel according to an embodiment of the present disclosure is schematically shown.

[0051] like Figure 2 As shown in Figure 1, when a train passes through a tunnel at high speed, the pressure fluctuation outside the train changes with time. When the train passes through the tunnel, the pressure outside the train is mainly negative. In some cases, the extreme negative pressure is >4000Pa. Figure 2 The drastic pressure fluctuations shown in the figure. If the valve is kept open, this pressure fluctuation will quickly spread to the train, causing excessive pressure changes in the train and causing passengers to feel uncomfortable with ear pressure. Therefore, the valve is kept closed while the train passes through the tunnel.

[0052] Figure 3 A flow chart of a train valve control method according to an embodiment of the present disclosure is schematically shown.

[0053] like Figure 3 As shown, the train valve control method of this embodiment includes operations S310 to S330.

[0054] In operation S310, on-board information and a mapping relationship corresponding to the train compartment are obtained, wherein the on-board information includes operation information, pressure information, and air quality information, and the mapping relationship represents the correspondence between the valve control information of the train compartment and the on-board information.

[0055] According to an embodiment of the present disclosure, the on-board information may be information obtained in real time by a monitoring device arranged on the train, including operation information, pressure information and air quality information, and the operation information includes speed information, line information and time information. The pressure information may include pressure difference information and pressure change rate. The air quality information may include carbon dioxide concentration information and oxygen content information. The mapping relationship may include a first mapping relationship and a second mapping relationship, the first mapping relationship may characterize the correspondence between the pressure difference information and the valve control information, and the second mapping relationship may characterize the correspondence between the speed information and the valve control information. It is understandable that the mapping relationship and the obtained on-board information may be stored in the server for subsequent analysis and processing.

[0056] In operation S320, the operation information, the pressure information, and the air quality information are input into the prediction model, and a prediction result is output, wherein the prediction result includes a pressure prediction result and an air quality prediction result.

[0057] According to an embodiment of the present disclosure, the prediction model may be an estimation model suitable for data sparse conditions. The pressure prediction result may characterize the pressure change rate in the vehicle and the pressure difference information between the inside and outside of the vehicle obtained by inputting the operation information, pressure information and air quality information into the prediction model. The air quality prediction result may characterize the carbon dioxide concentration information and oxygen content information in the vehicle compartment output by the prediction model.

[0058] In operation S330 , the state of the valve is switched or maintained based on the prediction result and the mapping relationship.

[0059] According to an embodiment of the present disclosure, the valve may represent a control valve for controlling the exchange of air inside the vehicle and outside the vehicle in the air conditioning system of a high-speed train. When the prediction result represents that the pressure prediction result and the air quality prediction result exceed a preset threshold, the train valve is switched or maintained based on the mapping relationship to meet the pressure difference and air quality requirements in the vehicle compartment.

[0060] According to the embodiments of the present disclosure, by inputting the real-time operation information, pressure information and air quality information into the prediction model, the pressure prediction results and air quality prediction results inside the car are obtained in real time, so that the valve state can be flexibly switched or maintained based on the mapping relationship between the valve control information and the on-board information. Since the prediction result is generated in real time based on the prediction model for different operation information, pressure information and air quality information, it satisfies the control of pressure fluctuations under different speed levels and different line conditions (such as complex tunnel groups in mountainous areas), avoids the situation where the passenger comfort is affected by excessive pressure fluctuations in the car or poor air quality, and further improves the passenger riding experience.

[0061] According to an embodiment of the present disclosure, the method also includes: obtaining feedback information corresponding to the state, wherein the feedback information represents the difference result between the state and a preset state threshold; and when the difference result represents that the state is less than the preset state threshold, using a preset update strategy to update the prediction model to obtain a target prediction model.

[0062] According to the embodiments of the present disclosure, considering the different actual conditions in different lines, the prediction accuracy of the prediction model is difficult to meet the actual needs of different lines, so it is considered to update the prediction model through feedback information in the actual operation of the line to adapt to the actual conditions of more different lines to improve the prediction accuracy of the model. The feedback information can represent the difference between the monitoring information obtained in real time by the monitoring equipment in the actual operation of the train and the prediction result output by the prediction model, which can be represented by the feedback information obtained by the valve state during operation.

[0063] For example, in line L1, the prediction model predicts that the carbon dioxide concentration in the carriage of train A at time t is less than the preset concentration threshold, indicating that the valve should remain closed. However, the carbon dioxide concentration obtained by the actual monitoring equipment has exceeded the preset concentration threshold (for example, 2500ppm), and the valve needs to be opened to introduce fresh air at time t. The above situation shows that the prediction accuracy of the prediction model does not meet the preset accuracy requirements in line L1. The prediction model can be updated through the preset update strategy to obtain a target prediction model that meets the requirements of train A in line L1.

[0064] According to the embodiments of the present disclosure, the prediction model is optimized and updated by using the real-time monitoring data of the train during operation, and the model can be regularly adjusted and optimized to better adapt to different operating lines and changing operating environments. At the same time, updating the model can reduce the prediction deviation caused by the use of outdated data and improve the overall prediction accuracy of the model.

[0065] According to an embodiment of the present disclosure, a prediction model includes an initial interaction layer; updating the prediction model using a preset update strategy to obtain a target prediction model includes: updating the initial interaction layer using a preset update strategy to obtain a target interaction layer, wherein the target interaction layer includes a first interaction layer and a second interaction layer, and the second interaction layer includes multiple feature vector processing sublayers; and updating the prediction model based on the second interaction layer to obtain a target prediction model.

[0066] According to an embodiment of the present disclosure, the prediction model may be an estimation model suitable for processing sparse data, such as a factorization machine model (FM). The factorization machine model may include a feature input layer, an embedding layer, an initial interaction layer (factorization machine layer), and an output layer.

[0067] According to an embodiment of the present disclosure, a method for updating an initial interaction layer using a preset update strategy may include: updating the initial interaction layer to a target interaction layer including a first interaction layer and a second interaction layer based on a Wide and deep strategy, wherein the first interaction layer can be used to process intersection problems between features, and multiple inner product units are used to combine different feature vectors in pairs, thereby enhancing the feature combination capability of a shallow network; the second interaction layer can be based on a Wide and deep strategy, using multiple feature vector processing sublayers (e.g., a multilayer perceptron) to perform deep processing of all features, thereby causing the updated prediction model to include a target interaction layer.

[0068] It is understandable that based on the target prediction model, in the subsequent train operation process, the output data of the first interaction layer and the second interaction layer can be nonlinearly transformed to obtain the output prediction results, and the valves can be controlled in combination with the mapping relationship. Based on the needs of different lines and different operating environments, the target prediction model can be personalized and updated and optimized. It should be noted that the second interaction layer can also be called a hidden layer. The hidden layer can map the input information to a higher-dimensional feature space, and help the neural network better understand the information by learning and extracting the features of the input information. Each hidden layer can learn different levels of abstract features. As the number of hidden layers increases, the model can gradually learn more abstract and complex feature representations.

[0069] According to an embodiment of the present disclosure, the method also includes: classifying the operation information, pressure information and air quality information using preset classification rules to obtain multiple feature domain information; processing the multiple feature domain information to obtain processing results; and determining the target prediction result based on the mapping relationship and the processing result, the target prediction result characterizing the state of the valve in the target time period.

[0070] According to the embodiments of the present disclosure, the monitoring information acquired in real time is input into the feature input layer of the target prediction model, and the information is classified according to the category features corresponding to the monitoring information and the preset classification rules to obtain feature domain information, which can also be called a one-hot vector; thereby, the one-hot vector is passed through the embedding layer and the target interaction layer to obtain a processing result, and then the state of the valve in the target time period is determined based on the mapping relationship. The embedding layer can obtain dense features by processing and converting the sparse information, and the output result of the target interaction layer can be converted nonlinearly to obtain a processing result, which can characterize the pressure prediction result and air quality prediction result of the train at the target time / time period obtained by the target prediction model.

[0071] In a feasible embodiment, a method for classifying operation information, pressure information, and air quality information based on preset classification rules to obtain feature domain information may include: setting one or more thresholds for each category, and dividing numerical data (such as pressure, air quality indicators) into different categories. For example, air quality can be set to levels such as "good", "medium", and "poor" according to carbon dioxide concentration; and generating a corresponding one-hot vector based on the classification result. For example, if air quality is divided into three categories, a vector of three elements can be generated, where the corresponding category is 1 and the rest are 0.

[0072] In another feasible embodiment, classification rules can be formulated based on domain knowledge or relevant experience, such as "normal operation", "abnormal operation" or "normal pressure", "too high pressure", etc.; these rules are applied according to real-time data and category labels are generated; one-hot vectors are generated according to the category labels, and each category corresponds to a vector position.

[0073] According to an embodiment of the present disclosure, multiple feature domain information is processed to obtain a processing result, including: mapping the multiple feature domain information to a low-dimensional continuous space to obtain multiple feature vectors; determining interactive feature information between the multiple feature vectors, wherein the interactive feature information includes first interactive information obtained using a first interactive layer and second interactive information obtained using a second interactive layer; and updating the first interactive information and the second interactive information based on a preset conversion rule to obtain a processing result.

[0074] According to an embodiment of the present disclosure, a method for mapping multiple feature domain information to a low-dimensional continuous space using an embedding layer to obtain multiple feature vectors may include: converting the feature domain information into a dense feature representation of a fixed size, mainly by reducing the dimension of sparse features, and calculating the weight matrix of the embedding layer to reduce the dimension, thereby obtaining multiple dense feature vectors. Compared with sparse features, dense feature vectors have higher information density and can retain more original information in a lower dimension.

[0075] For example, the feature domain information is numerically vectorized, and the first column is set to be the feature of speed greater than 200km / h, 1 represents yes, 0 represents no, and other features are processed in the same way to obtain [1,0,1,0,1,1,1]. It can be understood that two or more feature domain information (such as speed, altitude) can also be composed of a vector matrix.

[0076] Figure 4 The structural diagram of the target prediction model according to the embodiment of the present disclosure is schematically shown.

[0077] like Figure 4As shown, the target prediction model may include a feature input layer 410, an embedding layer 420, a target interaction layer 430 and an output layer 440, and the target interaction layer 430 may include a first interaction layer 431 and a second interaction layer 432. The interactive feature information may be based on the target interaction layer 430 processing multiple feature vectors, and based on the Wide and deep strategy, the first interaction layer (factor decomposition machine layer) 431 may be used to replace the wide part in the Wide and deep strategy to enhance the feature combination capability of the shallow network and obtain the first interactive information; the second interactive layer 432 maintains consistency with the Wide and deep strategy (i.e., the deep part), and uses multi-layer neurons (such as multi-layer perceptrons) to perform deep processing of all features to obtain the second interactive information; thereby, the output results of the target interaction layer (the first interactive information and the second interactive information) are combined and linearly transformed to obtain the transformation processing result.

[0078] According to the embodiments of the present disclosure, the prediction model is updated and optimized through the Wide and Deep strategy. The Wide part can effectively utilize the direct interaction of features, including sparse features, such as different train types, operating status, etc., to help the model quickly learn specific associations; the Deep part can capture more complex feature interactions through a deep neural network, which can deeply explore the potential relationship between features and improve the expression ability of the model. By combining the linear characteristics of the wide model and the nonlinear characteristics of the deep model, the monitoring information obtained in real time can be more fully understood, thereby improving the accuracy of the train pressure prediction results and the air quality prediction results.

[0079] According to an embodiment of the present disclosure, switching or maintaining the state of a valve based on prediction results and mapping relationships includes: when the prediction results and mapping relationships indicate switching the valve state for a target time period, switching the valve state based on preset switching rules, wherein the preset switching rules include at least one of car-by-car switching rules and time-by-time switching rules.

[0080] According to the embodiments of the present disclosure, taking into account different situations in the actual operation of the train (such as the operating speed, driving environment and the number of passengers), when it is necessary to switch the valve state, it can be switched by switching in time periods, switching in carriages or by a combination of time periods and carriages.

[0081] In a feasible embodiment, the opening and closing time of the valve is reasonably controlled according to the actual conditions of the train operation (speed, driving environment, pressure difference between inside and outside the carriage, etc.); thereby, by real-time monitoring of the pressure changes in the carriage, the valve state can be adjusted in time to cope with emergencies.

[0082] In another feasible embodiment, different valve opening strategies can be set according to the characteristics of different carriages (passenger capacity, purpose, etc.); for specific carriages (the head and tail cars of the EMU), avoid opening the valve during operation or arrange a suitable time to open the valve to reduce the impact of pressure waves on the driver; through pressure sensors and feedback mechanisms, pressure sensors can be installed inside each carriage to monitor internal pressure changes in real time; sensor data is fed back to the control system to form a closed-loop control and achieve dynamic adjustment.

[0083] Figure 5 A diagram of a train interior pressure wave protection system according to an embodiment of the present disclosure is schematically shown.

[0084] like Figure 5 As shown, in the train pressure wave protection system, an external air connection pipe 510 is used to connect to the external atmosphere, and the pressure change outside the car is detected through the pressure control unit 501; a control system control signal 520 is used to send instructions to the protection valve, and the protection valve includes an air conditioning pressure protection valve 5021 and a waste exhaust pressure protection valve 5022; the train air compressor 503 is used to control the air conditioning pressure protection valve 5021 and the waste exhaust pressure protection valve 5022 to realize the compressed air flow direction 530; the internal air connection pipe 540 can sense the pressure change inside the car through the pressure control unit 501. It can be understood that the number of pressure control units, protection valves and train air compressors can be determined according to actual conditions, and is not specifically limited here.

[0085] According to an embodiment of the present disclosure, the pressure information includes pressure difference information, the operation information includes speed information, the mapping relationship includes a first mapping relationship and a second mapping relationship, the first mapping relationship represents the correspondence between the pressure difference information and the valve control information, and the second mapping relationship represents the correspondence between the speed information and the valve control information.

[0086] According to an embodiment of the present disclosure, the first mapping relationship may represent the correspondence between the pressure difference information and the valve control information, and the pressure difference information may represent the information of the continuous internal and external pressure difference ADP(t) of the compartment. The correspondence (control logic) between the pressure difference information and the valve control information may include: valve closing logic, valve opening logic and pressure relief logic. The second mapping relationship may represent the correspondence between the speed information and the valve control information.

[0087] It is understandable that the valve closing logic, valve opening logic and pressure relief logic can be limited according to actual conditions and specific needs, and are not specifically limited here.

[0088] It should be noted that in order to reduce the impact of the pressure change rate in the car on the ride comfort, the present disclosure optimizes the pressure change rate in the car by defining the control conditions of the valve opening logic in the model. For example, the pressure change rate in the car within 1s is less than 300Pas / s, the pressure change rate in the car within 3s is less than 500Pa / 3s, and the carbon dioxide concentration in the car is less than 1000ppm as the control target, and the mapping relationship is updated. At the same time, the valve opening and closing can be realized by time-sharing and car-sharing in combination with the number of passengers in different compartments and the actual situation.

[0089] According to the embodiments of the present disclosure, the valve state is controlled through mapping relationships and prediction results, and the pressure change rate is controlled by opening the valves in different time periods and compartments. This can reduce the impact of pressure waves on passengers, effectively slow down the propagation of pressure waves, improve passenger comfort, and reduce dizziness or discomfort. It also helps to maintain a stable state inside the compartment, reduce the possibility of failures, and enhance operational safety.

[0090] According to an embodiment of the present disclosure, the prediction model is trained by the following operations: obtaining sample initial on-board information corresponding to a sample train, wherein the sample initial on-board information includes initial operation information, initial pressure information, and initial air quality information; processing the sample initial on-board information to obtain sample on-board information, wherein the amount of information in the sample on-board information is greater than the amount of information in the sample initial on-board information; and training an initial prediction model using the sample on-board information to obtain a prediction model, wherein the sample on-board information includes a label corresponding to the sample on-board information.

[0091] According to the embodiments of the present disclosure, the sample initial vehicle information can represent the vehicle information in the historical time period obtained by the monitoring device. It is understandable that the sample initial vehicle information can be obtained by monitoring the vehicle information in the historical time period and storing it in the target information library for actual needs. Considering that in actual operation, the amount of abnormal information is small and it is difficult to meet the training requirements of the initial prediction model, new training data can be generated by processing the sample initial vehicle information to meet the model training requirements.

[0092] According to an embodiment of the present disclosure, an initial prediction model is trained using sample vehicle information to obtain a prediction model, including: using the initial prediction model to classify the sample vehicle information to obtain multiple sample feature domain information; processing the multiple sample feature domain information to obtain a sample processing result; determining a sample prediction result based on the sample processing result and a sample mapping relationship; determining a difference between the sample prediction result and a verification result; and when the difference meets a preset threshold, determining the initial prediction model as the prediction model.

[0093] According to an embodiment of the present disclosure, the sample vehicle information may include a training set and a validation set. The initial prediction model may be a factor decomposition machine model, including a feature input layer, an embedding layer, an initial interaction layer, an output layer, and a loss layer. The difference can characterize the difference between the prediction result of the trained initial prediction model (intermediate prediction model) and the verification result in the sample vehicle information, which can be determined by the loss layer (loss function) to guide the training process of the model. It is understandable that the prediction model is determined when the prediction value of the intermediate prediction model meets the preset accuracy threshold. It should be noted that the loss function can be determined according to actual needs, such as mean square error, cross entropy loss, regularization loss or logarithmic loss, which is not limited here.

[0094] In a feasible embodiment, the loss function of the factor decomposition machine model can also be determined by Hinge loss. The method may include: in operation S401, the model outputs a prediction value based on a defined bias term, a linear weight, and a factor vector; in operation S402, the loss function is calculated using Hinge loss. In order to prevent overfitting, a regularization term may be added to the loss function; in operation S403, the loss function is derived for use in the optimization process; in operation S404, gradient descent or other optimization algorithms (such as Adam, AdaGrad, etc.) are used to update the model parameters; and the above operations S403 to S404 are repeated until the loss converges or the set number of iterations is reached.

[0095] According to an embodiment of the present disclosure, multiple sample feature domain information is processed to obtain a sample processing result, including: mapping the multiple sample feature domain information to a low-dimensional continuous space to obtain multiple sample feature information; determining sample interaction feature information between the multiple sample feature information; and performing nonlinear transformation on the sample interaction feature information to obtain a sample processing result.

[0096] According to the embodiments of the present disclosure, discrete features can be mapped to a low-dimensional continuous space through an embedding layer, and each feature corresponds to an embedding vector. Through the embedding of features, the interaction relationship between discrete features can be effectively determined; thereby, the inner product of the embedded features is calculated through the interaction layer to determine the quadratic interaction between the features, and sample interaction feature information is obtained, and then the sample interaction feature information is nonlinearly transformed to obtain the sample processing result.

[0097] Figure 6 The structural diagram of the initial prediction model according to an embodiment of the present disclosure is schematically shown.

[0098] like Figure 6As shown, the initial prediction model may include a feature input layer 610, an embedding layer 620, an initial interaction layer 630, and an output layer 640. The input layer 610 may be used to receive sparse feature information and pass it to subsequent layers in the form of vectors; the embedding layer 620 may map discrete features into a low-dimensional continuous space, and by embedding features, the interactive relationship between features may be effectively captured; the initial interaction layer 630 may capture the secondary interaction between features by calculating the inner product of feature embeddings, and utilizing interactive features may help improve the performance of the model; the output layer 640 may map the output result of the initial interaction layer 630 to the final predicted value.

[0099] According to an embodiment of the present disclosure, the sample initial vehicle information is processed to obtain the sample vehicle information, including: using a preset processing strategy to process the initial abnormal information in the sample initial vehicle information to obtain abnormal information; and updating the sample initial vehicle information based on the abnormal information to obtain the sample vehicle information.

[0100] According to an embodiment of the present disclosure, a preset processing strategy can be used to process the initial abnormal information to obtain new abnormal information, and the preset processing strategy includes a random noise strategy and a translation strategy. The initial abnormal information can characterize the abnormal operation information, abnormal pressure information and abnormal air quality information in the train on-board data in the historical time period. Generating new abnormal information through the initial abnormal information can make up for the lack of training data to meet the needs of model training.

[0101] In a feasible embodiment, new sample vehicle information can also be generated through a generative adversarial network (CGAN), including: obtaining sample vehicle information including multiple features, and assigning a label to each sample to identify its features, such as running speed, running route, in-vehicle pressure difference, maximum pressure difference and carbon dioxide concentration, etc.; using the CGAN generator to take random noise vectors and conditional information (labels) as input to generate feature data similar to real samples, using the discriminator to receive real samples or generated samples and conditional labels, and judge their authenticity, and judge whether the samples are real or generated through binary classification output; using the trained generator to input specific noise vectors and target feature labels to generate new samples that meet the requirements.

[0102] According to the embodiments of the present disclosure, a preset processing strategy is used to process the sample vehicle information obtained during operation. In response to the situation where there are insufficient samples of certain categories (such as abnormal information during operation), more samples of that category can be generated to help the model learn better. By increasing the amount and diversity of training data, the generalization ability of the model is improved, thereby improving the final prediction effect of the model.

[0103] According to an embodiment of the present disclosure, the preset processing strategy includes a random noise strategy; the initial abnormal information in the sample initial vehicle information is processed using the preset processing strategy to obtain abnormal information, including: determining the noise type and noise parameters corresponding to the initial abnormal information; generating noise information based on the noise type and noise parameters; and generating abnormal information using the noise information and the initial abnormal information.

[0104] According to an embodiment of the present disclosure, the noise type may include Gaussian noise or balanced noise, and the noise parameter may characterize the standard deviation or range of the noise. The method of obtaining abnormal information using a random noise strategy includes: creating a random noise vector, which can be generated using a normal distribution or a uniform distribution, and the dimension of the noise should match the target feature space; thereby combining the random noise with the existing features according to the type of feature to be generated. For example, the random noise can be weighted and summed with the existing features, or new features can be formed through other operations (such as multiplication, concatenation, etc.); a variational autoencoder can be used to receive random noise and existing features and output new samples (abnormal information).

[0105] According to an embodiment of the present disclosure, the preset processing strategy also includes a translation strategy; using the preset processing strategy to process the initial abnormal information in the sample initial vehicle information to obtain the abnormal information, it also includes: determining a translation value corresponding to the initial abnormal information, wherein the translation value represents the movement amplitude of the initial abnormal information over the target historical time period; and translating the initial abnormal information based on the movement amplitude to obtain the abnormal information.

[0106] According to an embodiment of the present disclosure, a translation strategy can be characterized by using a time axis translation to generate abnormal information, including: determining how many units to translate on the time axis, which can be a positive value or a negative value; moving the time series data forward or backward according to the selected translation amount; for data out of range, a missing value filling processing method can be used to ultimately generate abnormal data.

[0107] It is understandable that the random noise strategy and the translation strategy can also be combined to process the initial abnormal data, first by adding noise and then performing time translation to obtain the abnormal data.

[0108] Based on the above train valve control method, the present disclosure also provides a train valve control device. Figure 7 The device is described in detail.

[0109] Figure 7 The structural block diagram of the train valve control device according to an embodiment of the present disclosure is schematically shown.

[0110] like Figure 7As shown, the train valve control device of this embodiment includes an information acquisition module 710 , a prediction result output module 720 and a switching or maintaining module 730 .

[0111] The information acquisition module 710 is used to obtain the on-board information and mapping relationship corresponding to the train compartment, wherein the on-board information includes operation information, pressure information and air quality information, and the mapping relationship represents the corresponding relationship between the valve control information of the train compartment and the on-board information. In one embodiment, the information acquisition module 710 can be used to perform the operation S310 described above, which will not be repeated here.

[0112] The prediction result output module 720 is used to input the operation information, pressure information and air quality information into the prediction model and output the prediction result, wherein the prediction result includes the pressure prediction result and the air quality prediction result. In one embodiment, the prediction result output module 720 can be used to perform the operation S320 described above, which will not be repeated here.

[0113] The switching or maintaining module 730 is used to switch or maintain the state of the valve based on the prediction result and the mapping relationship. In one embodiment, the switching or maintaining module 730 can be used to perform the operation S330 described above, which will not be described in detail here.

[0114] According to the embodiments of the present disclosure, through the information acquisition module 710, the prediction result output module 720 and the switching or maintaining module 730 in the train valve control device, by inputting the real-time acquired operation information, pressure information and air quality information into the prediction model, the pressure prediction result and the air quality prediction result inside the car are obtained in real time, so as to flexibly switch or maintain the valve state based on the mapping relationship between the valve control information and the on-board information. Since the prediction result is generated in real time based on the prediction model for different operation information, pressure information and air quality information, it satisfies the control of pressure fluctuations under different speed levels and different line conditions (such as complex tunnel groups in mountains), avoids the situation where the passenger comfort is affected by excessive pressure fluctuations in the car or poor air quality, and further improves the passenger riding experience.

[0115] According to an embodiment of the present disclosure, the device further includes: a feedback information acquisition module and a prediction model updating module.

[0116] The feedback information acquisition module is used to acquire feedback information corresponding to the state, wherein the feedback information represents the difference result between the state and a preset state threshold.

[0117] The prediction model updating module is used to update the prediction model using a preset updating strategy to obtain a target prediction model when the state represented by the difference result is less than a preset state threshold.

[0118] According to an embodiment of the present disclosure, the prediction model includes an initial interaction layer; the prediction model update module includes: an initial interaction layer update submodule and a model update submodule.

[0119] The initial interaction layer updating submodule is used to update the initial interaction layer using a preset updating strategy to obtain a target interaction layer, wherein the target interaction layer includes a first interaction layer and a second interaction layer, and the second interaction layer includes a plurality of feature vector processing sublayers.

[0120] The model updating submodule is used to update the prediction model based on the second interaction layer to obtain a target prediction model.

[0121] According to an embodiment of the present disclosure, the device further includes: an information classification module, an information processing module and a prediction result determination module.

[0122] The information classification module is used to classify the operation information, pressure information and air quality information using preset classification rules to obtain multiple feature domain information.

[0123] The information processing module is used to process multiple feature domain information to obtain processing results.

[0124] The prediction result determination module is used to determine a target prediction result based on the mapping relationship and the processing result, and the target prediction result represents the state of the valve in the target time period.

[0125] According to an embodiment of the present disclosure, the information processing module includes: a mapping submodule, an interaction feature information determination submodule and an information update processing submodule.

[0126] The mapping submodule is used to map multiple feature domain information to a low-dimensional continuous space to obtain multiple feature vectors.

[0127] The interactive feature information determination submodule is used to determine the interactive feature information between multiple feature vectors, wherein the interactive feature information includes first interactive information obtained by using the first interactive layer and second interactive information obtained by using the second interactive layer.

[0128] The information update processing submodule is used to update the first interaction information and the second interaction information based on a preset conversion rule to obtain a processing result.

[0129] According to an embodiment of the present disclosure, the switching or maintaining module includes: a state switching submodule, which is used to switch the state of the valve based on preset switching rules when the prediction result and the mapping relationship indicate that the valve state should be switched in the target time period, wherein the preset switching rules include at least one of the car switching rules and the time period switching rules.

[0130] According to an embodiment of the present disclosure, the pressure information includes pressure difference information, the operation information includes speed information, the mapping relationship includes a first mapping relationship and a second mapping relationship, the first mapping relationship represents the correspondence between the pressure difference information and the valve control information, and the second mapping relationship represents the correspondence between the speed information and the valve control information.

[0131] According to an embodiment of the present disclosure, the prediction model is trained by the following operations: obtaining sample initial on-board information corresponding to a sample train, wherein the sample initial on-board information includes initial operation information, initial pressure information, and initial air quality information; processing the sample initial on-board information to obtain sample on-board information, wherein the amount of information in the sample on-board information is greater than the amount of information in the sample initial on-board information; and training an initial prediction model using the sample on-board information to obtain a prediction model, wherein the sample on-board information includes a label corresponding to the sample on-board information.

[0132] According to an embodiment of the present disclosure, an initial prediction model is trained using sample vehicle information to obtain a prediction model, including: using the initial prediction model to classify the sample vehicle information to obtain multiple sample feature domain information; processing the multiple sample feature domain information to obtain a sample processing result; determining a sample prediction result based on the sample processing result and a sample mapping relationship; determining a difference between the sample prediction result and a verification result; and when the difference meets a preset threshold, determining the initial prediction model as the prediction model.

[0133] According to an embodiment of the present disclosure, multiple sample feature domain information is processed to obtain a sample processing result, including: mapping the multiple sample feature domain information to a low-dimensional continuous space to obtain multiple sample feature information; determining sample interaction feature information between the multiple sample feature information; and performing nonlinear transformation on the sample interaction feature information to obtain a sample processing result.

[0134] According to an embodiment of the present disclosure, the sample initial vehicle information is processed to obtain the sample vehicle information, including: using a preset processing strategy to process the initial abnormal information in the sample initial vehicle information to obtain abnormal information; and updating the sample initial vehicle information based on the abnormal information to obtain the sample vehicle information.

[0135] According to an embodiment of the present disclosure, the preset processing strategy includes a random noise strategy; the initial abnormal information in the sample initial vehicle information is processed using the preset processing strategy to obtain abnormal information, including: determining the noise type and noise parameters corresponding to the initial abnormal information; generating noise information based on the noise type and noise parameters; and generating abnormal information using the noise information and the initial abnormal information.

[0136] According to an embodiment of the present disclosure, the preset processing strategy also includes a translation strategy; using the preset processing strategy to process the initial abnormal information in the sample initial vehicle information to obtain the abnormal information, it also includes: determining a translation value corresponding to the initial abnormal information, wherein the translation value represents the movement amplitude of the initial abnormal information over the target historical time period; and translating the initial abnormal information based on the movement amplitude to obtain the abnormal information.

[0137] According to an embodiment of the present disclosure, any multiple modules of the information acquisition module 710, the prediction result output module 720 and the switching or holding module 730 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the information acquisition module 710, the prediction result output module 720 and the switching or holding module 730 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in any appropriate combination of any of them. Alternatively, at least one of the information acquisition module 710, the prediction result output module 720 and the switching or holding module 730 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be performed.

[0138] Based on the train valve control device, the present disclosure also provides a train, including: a train car; a monitoring device for monitoring the on-board information of the train car; a valve such as a train valve control method, for controlling the exchange between the air inside the train car and the outside air; and a train valve control device.

[0139] It should be noted that the installation locations of the monitoring equipment and valves in the train can be determined according to actual conditions and are not specifically limited here.

[0140] Figure 8 A block diagram of an electronic device suitable for implementing a train valve control method according to an embodiment of the present disclosure is schematically shown.

[0141] like Figure 8As shown, the electronic device according to the embodiment of the present disclosure includes a processor 801, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 802 or the program loaded from the storage part 808 to the random access memory (RAM) 803. The processor 801 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 801 may also include an onboard memory for caching purposes. The processor 801 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present disclosure.

[0142] In RAM 803, various programs and data required for the operation of electronic device 800 are stored. Processor 801, ROM 802 and RAM 803 are connected to each other via bus 804. Processor 801 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 802 and / or RAM 803. It should be noted that the program can also be stored in one or more memories other than ROM 802 and RAM 803. Processor 801 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0143] According to an embodiment of the present disclosure, the electronic device 800 may further include an input / output (I / O) interface 805, which is also connected to the bus 804. The electronic device 800 may further include one or more of the following components connected to the input / output (I / O) interface 805: an input portion 806 including a keyboard, a mouse, etc.; an output portion 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 808 including a hard disk, etc.; and a communication portion 809 including a network interface card such as a LAN card, a modem, etc. The communication portion 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the input / output (I / O) interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed, so that the computer program read therefrom is installed into the storage portion 808 as needed.

[0144] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0145] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 802 and / or RAM 803 described above and / or one or more memories other than ROM 802 and RAM 803.

[0146] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the train valve control method provided by the embodiment of the present disclosure.

[0147] The above functions defined in the system / device of the embodiment of the present disclosure are executed when the computer program is executed by the processor 801. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0148] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 809, and / or installed from a removable medium 811. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0149] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 809, and / or installed from the removable medium 811. When the computer program is executed by the processor 801, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0150] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0151] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0152] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0153] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A train valve control method, comprising: Acquire on-board information and a mapping relationship corresponding to the train compartment, wherein the on-board information includes operation information, pressure information, and air quality information, and the mapping relationship represents a corresponding relationship between the valve control information of the train compartment and the on-board information; Inputting the operation information, the pressure information and the air quality information into a prediction model, and outputting a prediction result, wherein the prediction result includes a pressure prediction result and an air quality prediction result; and Based on the prediction result and the mapping relationship, the state of the valve is switched or maintained.

2. The method according to claim 1, further comprising: Acquiring feedback information corresponding to the state, wherein the feedback information represents a difference result between the state and a preset state threshold; and When the difference result indicates that the state is less than the preset state threshold, the prediction model is updated using a preset update strategy to obtain a target prediction model.

3. The method according to claim 2, wherein: The prediction model includes an initial interaction layer; The prediction model is updated using a preset update strategy to obtain a target prediction model, including: Using the preset update strategy to update the initial interaction layer to obtain a target interaction layer, wherein the target interaction layer includes a first interaction layer and a second interaction layer, and the second interaction layer includes a plurality of feature vector processing sublayers; and The prediction model is updated based on the second interaction layer to obtain the target prediction model.

4. The method according to claim 3, further comprising: Classifying the operation information, the pressure information, and the air quality information using a preset classification rule to obtain a plurality of feature domain information; Processing the plurality of feature domain information to obtain a processing result; as well as Based on the mapping relationship and the processing result, a target prediction result is determined, where the target prediction result represents the state of the valve in a target time period.

5. The method according to claim 4, wherein: Processing the plurality of feature domain information to obtain processing results includes: Mapping the plurality of feature domain information to a low-dimensional continuous space to obtain a plurality of feature vectors; Determining interactive feature information between the plurality of feature vectors, wherein the interactive feature information includes first interactive information obtained using the first interactive layer and second interactive information obtained using the second interactive layer; and The first interaction information and the second interaction information are updated based on a preset conversion rule to obtain the processing result.

6. The method according to claim 1, wherein: Switching or maintaining the state of the valve based on the prediction result and the mapping relationship includes: When the prediction result and the mapping relationship indicate that the valve state is switched in the target time period, the valve state is switched based on a preset switching rule, wherein the preset switching rule includes at least one of a car switching rule and a time period switching rule.

7. The method according to claim 1, wherein: The pressure information includes pressure difference information, the operation information includes speed information, the mapping relationship includes a first mapping relationship and a second mapping relationship, the first mapping relationship represents the correspondence between the pressure difference information and the valve control information, and the second mapping relationship represents the correspondence between the speed information and the valve control information.

8. The method according to claim 1, wherein: The prediction model is trained by the following operations: Acquire sample initial vehicle-mounted information corresponding to the sample train, wherein the sample initial vehicle-mounted information includes initial operation information, initial pressure information, and initial air quality information; Processing the sample initial vehicle information to obtain sample vehicle information, wherein the amount of information in the sample vehicle information is greater than the amount of information in the sample initial vehicle information; and The initial prediction model is trained using the sample vehicle information to obtain the prediction model, wherein the sample vehicle information includes a label corresponding to the sample vehicle information.

9. The method according to claim 8, wherein: The initial prediction model is trained using the sample vehicle information to obtain the prediction model, including: Using the initial prediction model to classify the sample vehicle information, to obtain a plurality of sample feature domain information; Processing the plurality of sample feature domain information to obtain a sample processing result; Determine a sample prediction result based on the sample processing result and the sample mapping relationship; Determining the difference between the sample prediction result and the verification result; and When the difference satisfies a preset threshold, the initial prediction model is determined as the prediction model.

10. The method according to claim 9, wherein: Processing the plurality of sample feature domain information to obtain sample processing results includes: Mapping the plurality of sample feature domain information to a low-dimensional continuous space to obtain a plurality of sample feature information; Determining sample interaction feature information between a plurality of the sample feature information; and Nonlinear transformation is performed on the sample interaction feature information to obtain the sample processing result.

11. The method according to claim 8, wherein: The sample initial vehicle information is processed to obtain sample vehicle information, including: Using a preset processing strategy to process the initial abnormal information in the sample initial vehicle information to obtain abnormal information; and The sample initial vehicle-mounted information is updated based on the abnormal information to obtain the sample vehicle-mounted information.

12. The method according to claim 11, wherein: The preset processing strategies include random noise strategy; The initial abnormal information in the sample initial vehicle information is processed using a preset processing strategy to obtain abnormal information, including: Determining a noise type and a noise parameter corresponding to the initial abnormal information; generating noise information based on the noise type and the noise parameter; and The abnormality information is generated by using the noise information and initial abnormality information.

13. The method according to claim 12, wherein: The preset processing strategies also include translation strategies; The initial abnormal information in the sample initial vehicle information is processed using a preset processing strategy to obtain abnormal information, and further includes: Determining a translation value corresponding to the initial abnormal information, wherein the translation value represents a movement amplitude of the initial abnormal information over a target historical time period; and The initial abnormal information is translated based on the movement amplitude to obtain the abnormal information.

14. A train valve control device, comprising: An information acquisition module, used to acquire on-board information and a mapping relationship corresponding to a train compartment, wherein the on-board information includes operation information, pressure information and air quality information, and the mapping relationship represents a corresponding relationship between the valve control information of the train compartment and the on-board information; a prediction result output module, configured to input the operation information, the pressure information, and the air quality information into a prediction model, and output a prediction result, wherein the prediction result includes a pressure prediction result and an air quality prediction result; and The switching or maintaining module is used to switch or maintain the state of the valve based on the prediction result and the mapping relationship.

15. A train comprising: Train carriages; Monitoring equipment, used to monitor the onboard information of the train carriage; A valve as claimed in any one of claims 1 to 13, for controlling the exchange of air between the interior of the train compartment and the outside air; and The train valve control device as claimed in claim 14.

16. An electronic device, comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 13.

17. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the method according to any one of claims 1 to 13.

18. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 13 is implemented.

Images