CN114386715A

CN114386715A - Prediction method, system, equipment and medium for main braking air path pressure leakage

Info

Publication number: CN114386715A
Application number: CN202210121767.4A
Authority: CN
Inventors: 郭双全; 许伟; 谈宏志
Original assignee: Shanghai Electric Group Corp
Current assignee: Shanghai Electric Group Corp
Priority date: 2022-02-09
Filing date: 2022-02-09
Publication date: 2022-04-22

Abstract

The invention discloses a method, a system, equipment and a medium for predicting pressure leakage of a main braking air circuit, wherein the prediction method comprises the following steps: acquiring running state data of a target subway train in a target time period; screening parking state data corresponding to the longest parking time period of the target subway train from the running state data according to the parking confirmation parameters; calculating and obtaining effective data of the pressure of the main air path in the longest parking time period based on the parking state data; inputting the effective data of the main air path pressure into the leakage model, and outputting the leakage state of the main air path pressure of the braking of the target subway train in the target time period. According to the method, effective data of the pressure of the main air path is obtained through calculation under the condition that the accuracy of the sensor is limited; the reliability and effectiveness of pressure leakage prediction data are improved, and the efficiency and accuracy of main air path pressure prediction of the subway train are enhanced; the maintenance work of the main air path of the subway train is guided in time, and the maintenance cost is reduced.

Description

Prediction method, system, equipment and medium for main braking air path pressure leakage

Technical Field

The invention relates to the technical field of train brake main air path pressure detection, in particular to a method, a system, equipment and a medium for predicting brake main air path pressure leakage.

Background

In winter, the pipe joint has poor sealing performance due to expansion with heat and contraction with cold, and the pipeline has more air leakage and faults. In the traditional vehicle maintenance process, the detection mode of the air tightness leakage point mainly depends on methods such as ear listening, manual detection, soapy water detection and the like. However, this detection method is inefficient and a small leak is difficult to detect.

With the continuous and deep development of intelligent operation and maintenance level of subways and the rapid development of technologies such as sensors, embedded systems, network communication, big data and the like, subway trains generally transmit fault problems and key driving parameters occurring in the running process of vehicles to a ground data center through a vehicle-mounted data transmission device, so that the state of a train set is monitored in real time. However, due to the limitation of factors such as communication capacity and sensor cost, the types of gas path pressure signals transmitted by the vehicle are less, the adoption precision of pressure data is not high enough, and the accuracy of leakage condition prediction is low; and a sensor needs to be additionally arranged, so that the cost is high and the popularization is difficult.

Disclosure of Invention

The invention aims to overcome the defects of low efficiency and poor accuracy of the prediction of the pressure of the main braking air path in the prior art, and provides a method, a system, equipment and a medium for predicting the pressure leakage of the main braking air path.

The invention solves the technical problems through the following technical scheme:

in a first aspect, the present invention provides a method for predicting a pressure leakage of a main braking air path, the method comprising:

acquiring running state data of a target subway train in a target time period; the operating state data comprises parking confirmation parameters;

screening parking state data corresponding to the longest parking time period of the target subway train from the running state data according to the parking confirmation parameters;

calculating and obtaining the effective pressure data of the main air path in the longest parking time period based on the parking state data;

inputting the effective data of the main air path pressure into a leakage model, and outputting the leakage state of the main air path pressure of the braking of the target subway train in a target time period; the leakage model is a mathematical model constructed by a two-dimensional threshold value array and a two-dimensional attribute array, each threshold value in the two-dimensional threshold value array represents a retention time threshold value of the pressure of the main wind gas circuit at a set temperature and a set pressure, the attribute values of the two-dimensional attribute array comprise a true value and a false value, and the dimensionality of the two-dimensional threshold value array comprises the temperature and the pressure.

Preferably, the parking confirmation parameters comprise train speed, cab activation coefficient and air compressor working coefficient;

the step of screening out the parking state data corresponding to the longest parking time period of the target subway train from the running state data according to the parking confirmation parameters comprises the following steps:

intercepting at least one alternative parking time period from the target time period according to the train speed, the cab activation coefficient and the air compressor working coefficient;

or the like, or, alternatively,

the running state data further comprises a time parameter and a reference parameter, wherein the reference parameter comprises the pressure of a main braking air path; the parking confirmation parameters comprise train speed, cab activation coefficients and air compressor working coefficients;

intercepting at least one primary parking time period from the target time period according to the train speed, the cab activation coefficient and the air compressor working coefficient;

comparing every two adjacent air pressure change values in the main braking air path pressure with a first threshold value, and determining at least one alternative parking time period from the primary parking time period according to a comparison result;

screening out the longest time period from the alternative parking time periods as the longest parking time period;

and intercepting the parking state data corresponding to the longest parking time period from the running state data.

Preferably, the step of obtaining the main air path pressure effective data in the longest parking time period by calculation based on the parking state data comprises:

extracting a plurality of different air pressure values in the parking state data, deleting the maximum and minimum air pressure values, and generating an initial air pressure array;

eliminating the air pressure value corresponding to the sampling time interval larger than a second threshold value from the initial air pressure array, and generating a pressure holding set and overtime judgment information;

generating an alternative pressure holding sample set and pressure holding preset judgment information according to the pressure holding set and the parking state data;

comparing the number of samples in the alternative pressure maintaining sample set with a preset number;

if the number of the samples is not larger than the preset number, directly taking the alternative pressure maintaining sample set as an effective pressure maintaining sample set, and generating alternative pressure maintaining sample judgment information and main air path pressure effective data;

if the number of the samples is larger than the preset number, sorting the samples in the alternative pressure holding set according to the sequence of the pressure holding duration from large to small, screening to obtain an effective pressure holding sample set according with the preset number, and generating alternative pressure holding sample judgment information and main air path pressure effective data;

wherein the set of effective pressure-hold samples comprises at least one of a pressure-hold value, a hold temperature value, a pressure-hold duration, and a start time point.

Preferably, the step of generating a set of alternative pressure maintenance samples from the set of pressure maintenance samples and the parking state data comprises:

selecting at least one target pressure maintenance value from the set of pressure maintenance values;

comparing the difference value between the air pressure value and the target pressure holding value in the parking state data with a third threshold value, and then intercepting a plurality of candidate target pressure holding time periods from the longest parking time period according to the comparison result;

screening out the longest target pressure maintaining time period from the alternative target pressure time periods, and judging whether the longest target pressure maintaining time period meets a preset condition;

if so, generating the alternative pressure holding sample set according to the target pressure holding state data corresponding to the longest target pressure holding time period intercepted from the parking state data;

if not, then the step of selecting at least one target pressure maintenance value from the set of pressure maintenance values is performed until all of the pressure maintenance values in the set of pressure maintenance values have been traversed.

Preferably, the prediction method obtains the leakage model by training the following steps:

collecting pressure maintenance data of a plurality of normal main air circuit pressures of a subway train in a historical period;

and inputting the pressure maintenance data into an initial model for training to obtain the leakage model.

Preferably, the pressure maintenance data includes standard parameters, and the step of inputting the pressure maintenance data into an initial model for training to obtain the leakage model includes:

selecting the pressure maintenance data in which the standard parameter is within a set range as effective pressure maintenance data;

and correcting the pressure holding time threshold of the two-dimensional threshold array in the initial model based on the pressure holding time in the effective pressure holding data to generate the leakage model.

Preferably, after the operation state data of the target subway train in the target time period is obtained, the prediction method further includes:

preprocessing the running state data; the preprocessing comprises integrity detection and abnormal data correction of the running state data and generating preprocessing judgment information;

the main air path pressure effective data comprises the effective pressure holding sample set, the preprocessing judgment information, the overtime judgment information, the pressure holding preset judgment information and the alternative pressure holding sample judgment information.

In a second aspect, the present invention provides a system for predicting a pressure leak in a main brake air path, the system comprising:

the acquisition module is used for acquiring the running state data of the target subway train in a target time period; the operating state data comprises parking confirmation parameters;

the screening module is used for screening the parking state data corresponding to the longest parking time period of the target subway train from the running state data according to the parking confirmation parameters;

the calculation module is used for calculating and obtaining the effective pressure data of the main air path in the longest parking time period based on the parking state data;

the prediction module is used for inputting the effective data of the main air path pressure into a leakage model and outputting the leakage state of the main air path pressure of the braking of the target subway train in a target time interval; the leakage model is a mathematical model constructed by a two-dimensional threshold value array and a two-dimensional attribute array, each threshold value in the two-dimensional threshold value array represents a retention time threshold value of the pressure of the main wind gas circuit at a set temperature and a set pressure, the attribute values of the two-dimensional attribute array comprise a true value and a false value, and the dimensionality of the two-dimensional threshold value array comprises the temperature and the pressure.

Preferably, the parking confirmation parameters comprise train speed, cab activation coefficient and air compressor working coefficient; the screening module includes:

the first intercepting unit is used for intercepting at least one alternative parking time period from the target time period according to the train speed, the cab activation coefficient and the air compressor working coefficient;

or the like, or, alternatively,

the running state data further comprises a time parameter and a reference parameter, the reference parameter comprises the pressure of a main braking air circuit, and the parking confirmation parameter comprises the train speed, the cab activation coefficient and the working coefficient of an air compressor; the screening module includes:

the second intercepting unit is used for intercepting at least one primary parking time period from the target time period according to the train speed, the cab activation coefficient and the air compressor working coefficient;

the first comparison unit is used for comparing every two adjacent air pressure change values in the main braking air path pressure with a first threshold value and determining at least one alternative parking time period from the primary parking time period according to a comparison result;

the first screening unit is used for screening out the longest time period from the alternative parking time periods to be used as the longest parking time period;

and the third intercepting unit is used for intercepting the parking state data corresponding to the longest parking time period from the running state data.

Preferably, the calculation module includes:

the extraction unit is used for extracting a plurality of different air pressure values in the parking state data, deleting the maximum and minimum air pressure values and generating an initial air pressure array;

the removing unit is used for removing the air pressure value corresponding to the sampling time interval larger than a second threshold value from the initial air pressure array, and generating a pressure holding set and overtime judgment information;

the generating unit is used for generating an alternative pressure maintaining sample set and pressure maintaining preset judgment information according to the pressure maintaining set and the parking state data;

the second comparison unit is used for comparing the number of the samples in the alternative pressure maintaining sample set with a preset number;

the first processing unit is used for directly taking the alternative pressure maintaining sample set as an effective pressure maintaining sample set and generating alternative pressure maintaining sample judgment information and main air path pressure effective data if the number of the samples is not greater than the preset number;

the second processing unit is used for sorting the samples in the alternative pressure maintaining set according to the sequence of the pressure maintaining duration from large to small if the number of the samples is larger than the preset number, screening to obtain an effective pressure maintaining sample set which accords with the preset number, and generating alternative pressure maintaining sample judgment information and main air path pressure effective data;

Preferably, the generating unit is specifically configured to:

Preferably, the prediction system obtains the leakage model by training the following modules, including:

the system comprises an acquisition module, a storage module and a control module, wherein the acquisition module is used for acquiring pressure maintenance data of a plurality of normal main air circuit pressures of the subway train in a historical time period;

and the training module is used for inputting the pressure maintaining data into an initial model for training so as to obtain the leakage model.

Preferably, the pressure maintenance data includes standard parameters, and the training module includes:

a selection unit configured to select the pressure maintenance data in which the standard parameter is within a set range as effective pressure maintenance data;

and the correcting unit is used for correcting the pressure holding time threshold of the two-dimensional threshold array in the initial model based on the pressure holding time in the effective pressure holding data to generate the leakage model.

The prediction system further comprises:

the preprocessing module is used for preprocessing the running state data; the preprocessing comprises integrity detection and abnormal data correction of the running state data and generating preprocessing judgment information;

the main air path pressure effective data comprises the effective pressure maintaining sample set, the preprocessing judgment information, overtime judgment information, pressure maintaining preset judgment information and alternative pressure maintaining sample judgment information.

In a third aspect, the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the method for predicting the main braking air path pressure leakage according to the first aspect is implemented.

In a fourth aspect, the present invention provides a computer-readable storage medium having stored therein a computer program which, when executed by a processor, implements the method for predicting a main brake air path pressure leak of the first aspect.

The positive progress effects of the invention are as follows: the method comprises the steps of training to obtain a leakage model based on relevant parameters of a subway train braking main air path, calculating to obtain effective main air path pressure data based on running state data of the subway train, and inputting the effective main air path pressure data into the leakage model to predict the air pressure leakage condition of the train. According to the method, effective data of the pressure of the main air path is obtained through calculation under the condition that the accuracy of the sensor is limited; the reliability and effectiveness of pressure leakage prediction data are improved, and the efficiency and accuracy of main air path pressure prediction of the subway train are enhanced; the maintenance work of the main air path of the subway train is guided in time, and the maintenance cost is reduced.

Drawings

Fig. 1 is a flowchart of a method for predicting pressure leakage of a main brake air path according to embodiment 1 of the present invention.

Fig. 2 is a flowchart of step S3 of the method for predicting main brake air path pressure leakage according to embodiment 1 of the present invention.

Fig. 3 is a flowchart of step S4 of the method for predicting main brake air path pressure leakage according to embodiment 1 of the present invention.

Fig. 4 is a flowchart of step S43 of the method for predicting main brake air path pressure leakage according to embodiment 1 of the present invention.

Fig. 5 is a schematic block diagram of a system for predicting main brake air path pressure leakage according to embodiment 2 of the present invention.

Fig. 6 is a schematic diagram of a result of initial training of a leakage model of a prediction system for main braking air path pressure leakage in embodiment 2 of the present invention.

Fig. 7 is a schematic diagram of a result of the end of training of the leakage model of the system for predicting main brake air path pressure leakage in embodiment 2 of the present invention.

Fig. 8 is a schematic diagram of a hardware structure of an electronic device according to embodiment 3 of the present invention.

Detailed Description

The invention is further illustrated by the following examples, which are not intended to limit the scope of the invention.

Example 1

As shown in fig. 1, the present embodiment provides a method for predicting a pressure leakage of a main braking air path, which includes the following steps:

s1, acquiring the running state data of the target subway train in the target time period; the operating state data includes parking confirmation parameters.

S2, preprocessing the running state data; the preprocessing comprises the steps of carrying out integrity detection and abnormal data correction on the running state data; and generates preprocessing judgment information.

And S3, screening the parking state data corresponding to the longest parking time period of the target subway train from the running state data according to the parking confirmation parameters.

And S4, calculating and obtaining the effective data of the pressure of the main air path in the longest parking time period based on the parking state data.

And S5, inputting the effective data of the main air path pressure into the leakage model, and outputting the leakage state of the main air path pressure of the braking of the target subway train in the target time period. The leakage model is a mathematical model constructed by a two-dimensional threshold value array and a two-dimensional attribute array, the dimensionality of the two-dimensional threshold value array comprises temperature and pressure, each threshold value in the two-dimensional threshold value array represents a retention time threshold value of the pressure of a main air path under set temperature and set pressure, and the attribute value of the two-dimensional attribute array comprises a true value and a false value.

In step S1, the train data related to the train operation collected by the electronic device such as the sensor on the target subway train is received regularly. And acquiring running state data related to the main wind path pressure in a target time period (from 12 month 1 early morning in 2021 to 12 month 2 day 24 in 2021) from the driving data. The operation state data can comprise the current time, the pressure value of the main air path, the ambient temperature value outside the automobile, parking confirmation parameters and the like.

In step S2, the integrity of the running state data is first checked, and if the data is complete, it is further detected whether the data format meets a preset requirement (for example, a pkl or csv file), and if the data format meets the requirement, it is finally determined whether the data size meets the limit of the minimum data size. If any one of the above check results fails, a prompt message of data acquisition failure can be returned. And identifying abnormal data points corresponding to the main wind path pressure in the operation state data, such as data points of single mutation and data points exceeding a preset range (for example, the preset range is 7.5 bar-9.0 bar). And correcting the abnormal data points by means of neighbor value substitution, average value correction, default value correction and the like. If the correction fails, prompt information of the failure of the correction can be returned.

In step S3, a plurality of parking time periods are selected from the operating state data according to the parking confirmation parameter, and the longest parking time period is selected from the plurality of parking time periods. And further screening the parking state data corresponding to the longest parking time period from the running state data.

It can be understood that the parking confirmation parameter may be used as a first screening condition, and in order to ensure the stability of the pressure change, the pressure change interval of the main air path may be smaller than or equal to 0.1bar and may be used as a second screening condition, so as to determine the longest parking time period of the target subway train in the target time period.

In step S4, after the pressure value corresponding to the non-repetitive main air path pressure appearing in the parking state data is determined, the maximum pressure value and the minimum pressure value are removed, and finally invalid data such as the main air path pressure corresponding to the abnormal sampling time are screened out to generate a pressure holding value set.

Each time a target pressure maintenance value (e.g., 8.5bar) is selected from the set of pressure maintenance values, the parking state data is traversed, and at least one alternative pressure maintenance value is selected therefrom that is greater than or equal to 8.5bar and less than 8.6 bar. And further judging whether the sampling time periods corresponding to the target pressure holding value and the alternative pressure holding value are reasonable or not, and if so, selecting the longest sampling time period as the pressure holding duration corresponding to the main air circuit pressure holding value of 8.5bar of the target subway train in the target time period. And by analogy, selecting a next target pressure maintaining value (for example, 8.6bar) from the pressure maintaining set, determining the pressure maintaining duration corresponding to the main air gas circuit pressure maintaining value of 8.6bar of the target subway train in the target time interval according to the above method, and generating the effective main air gas circuit pressure data until all the pressure maintaining values in the pressure maintaining set traverse.

In step S5, the valid data of the pressure of the main air path is input into a leakage model trained in advance, and the result of the leakage state of the pressure of the target subway train in the target time interval is output. It can be understood that the data state result of the main air path pressure effective data can also be output. For example, the data state results may include data integrity analysis, preprocessing analysis, parking time presence analysis, timeout point presence analysis, pressure hold duration analysis, and the like.

In one implementation, as shown in fig. 2, the parking confirmation parameters include train speed, cab activation factor, and air compressor operating factor; step S3 specifically includes:

and S31, intercepting at least one alternative parking time period from the target time period according to the train speed, the cab activation coefficient and the air compressor working coefficient.

And S34, screening out the longest time period from the alternative parking time periods as the longest parking time period.

And S35, intercepting the parking state data corresponding to the longest parking time period from the running state data.

In step S31, the time slot in which the train speed is zero, the activation coefficients of the front and rear cabs are both zero, and the working coefficient of the air compressor is zero in each time slot in the running state data is taken as the alternative parking time slot of the target subway train.

In another practical solution, as shown in fig. 2, the operation status data further includes a time parameter and a reference parameter, and the reference parameter includes a main braking air path pressure; the parking confirmation parameters comprise the speed of the train, the activation coefficient of the cab and the working coefficient of the air compressor; step S3 specifically includes:

and S32, intercepting at least one primary parking time period from the target time period according to the train speed, the cab activation coefficient and the air compressor working coefficient.

And S33, comparing every two adjacent air pressure change values in the main braking air path pressure with the first threshold value, and determining at least one alternative parking time period from the primary parking time period according to the comparison result.

In step S32, the time slot in which the train speed is zero, the activation coefficients of the front and rear cabs are zero, and the working coefficient of the air compressor is zero in each time slot in the running state data is used as the initial parking time slot of the target subway train.

In the step S33-step S35, preset data processing is performed on the main brake air path pressure corresponding to the plurality of primary parking time periods. Specifically, after invalid primary parking time periods in which the change difference value of adjacent air pressure in the main braking air path pressure is greater than a first threshold value (0.1bar) are removed, the remaining primary parking time periods are used as alternative parking time periods, the longest parking time period is screened out from the alternative parking time periods, and the parking state data corresponding to the longest parking time period is further obtained. If all the initially selected parking time periods do not meet the conditions, prompt information of failure in obtaining the parking state data can be returned.

In an implementable scenario, as shown in fig. 3, step S4 specifically includes:

and S41, extracting a plurality of different air pressure values in the parking state data, deleting the maximum and minimum air pressure values, and generating an initial air pressure array.

And S42, eliminating the air pressure value corresponding to the sampling time interval larger than the second threshold value from the initial air pressure array, and generating a pressure holding set and overtime judgment information.

And S43, generating an alternative pressure maintaining sample set and pressure maintaining preset judgment information according to the pressure maintaining set and the parking state data.

And S44, comparing the number of the samples in the alternative pressure maintaining sample set with a preset number. If the number of samples is not greater than the predetermined number, go to step S45, and if the number of samples is greater than the predetermined number, go to step S46.

And S45, directly taking the alternative pressure maintaining sample set as an effective pressure maintaining sample set, and generating alternative pressure maintaining sample judgment information and main wind path pressure effective data.

And S46, sorting the samples in the alternative pressure maintaining set according to the sequence of the pressure maintaining duration from large to small, screening to obtain an effective pressure maintaining sample set which accords with the preset quantity, and generating alternative pressure maintaining sample judgment information and main air path pressure effective data.

Wherein the set of effective pressure-hold samples includes at least one of a pressure-hold value, a hold temperature value, a pressure-hold duration, and a start time point.

In step S41, the pressure values of all the non-repetitive main air path pressures in the parking state data are counted. The nonrepeating air pressure values were [8.6bar, 8.7bar, 8.8bar, 8.9bar ], the maximum value and the minimum value among the nonrepeating air pressure values were 8.9bar and 8.6bar, and the initial air pressure array was generated as [8.7bar, 8.8bar ].

In step S42, after two air pressure values with sampling time intervals greater than a second threshold (e.g., 10min) are removed from the initial air pressure array, a pressure hold set of [8.7bar, 8.8bar ] and timeout information are generated. The timeout information may include whether there is an abnormal data point using the time interval in the initial air pressure array, that is, an air pressure point of the timeout data.

In the steps S43 to S46, the number of samples in the candidate pressure holding set samples is compared with the preset number N, an effective pressure holding sample set is determined according to the comparison result, and then the main air path pressure effective data and the candidate pressure holding sample judgment information are generated according to the effective pressure holding sample set. The candidate pressure-holding sample judgment information includes information on whether the number of samples is too large, too small, or reasonable.

It should be noted that the valid data of the pressure of the main air path further includes preprocessing judgment information, the timeout judgment information, the pressure maintenance preset judgment information, and the alternative pressure maintenance sample judgment information.

In an implementable scenario, as shown in fig. 4, step S43 specifically includes:

s431, selecting at least one target pressure maintenance value from the set of pressure maintenance values.

S432, comparing the difference value between the air pressure value and the target pressure holding value in the parking state data with a third threshold value, and intercepting a plurality of candidate target pressure holding time periods from the longest parking time period according to the comparison result.

And S433, screening out the longest target pressure maintaining time period from the alternative target pressure time periods, and judging whether the longest target pressure maintaining time period meets a preset condition. If yes, step S434 is executed, and if not, step S431 is executed until all the pressure holding values in the pressure holding value set are traversed.

And S434, generating a candidate pressure holding sample set according to the target pressure holding state data corresponding to the longest target pressure holding period intercepted from the parking state data.

A target pressure maintenance value (e.g., 8.7bar) may be selected from the pressure maintenance set at a time in step S431, and a plurality of candidate target pressure maintenance values that are greater than or equal to 8.7bar and do not exceed one sensitivity unit (a third threshold, e.g., 0.1bar) are screened from the parked state data in step S432.

In step S433, after the longest target pressure holding time period is screened out in each candidate target pressure holding time period, it is determined whether the sampling time corresponding to the longest target pressure holding time period starts at the starting parking time of the target subway train or ends at the ending parking time of the target subway train. If so, rejecting the longest target pressure holding time period, if not, continuously judging whether the length of the pressure holding time period corresponding to the longest target pressure holding time period is less than 1 minute, if so, rejecting the longest target pressure holding time period, and if not, retaining the longest target pressure holding time period.

In one embodiment, the prediction method is trained to obtain the leakage model by the following steps:

the method comprises the steps of collecting pressure maintenance data of a plurality of normal main air circuit pressures of the subway train in a historical time period.

The pressure maintenance data is input into the initial model for training to obtain a leak model.

Specifically, pressure maintaining data of the main air circuit pressure of a certain subway train in a healthy state (no leakage or small reasonable leakage) during the period from 12/21/2018/21/2019/2/28/2019/6/29/2019/7/9/2019 are collected, and the pressure maintaining data of the two periods are input into an initial model for training by taking days as units. It will be appreciated that the initial state threshold arrays for this initial model are all set to 0, and the attribute arrays are all set to false.

The pressure maintenance data including standard parameters, the step of inputting the pressure maintenance data into an initial model for training to obtain a leakage model comprising:

selecting pressure maintenance data with standard parameters within a set range as effective pressure maintenance data;

and correcting the pressure holding time threshold of the two-dimensional threshold array in the initial model based on the pressure holding time in the effective pressure holding data to generate a leakage model.

Specifically, under the condition that the standard parameters can include temperature and pressure, the temperature change range is preset to be-10-50 ℃, the interval is 1 ℃, the pressure change range is set to be 7.5-9bar, and the interval is 0.1 bar; and taking the data which simultaneously meets the standard parameter setting conditions in the pressure maintaining data as effective pressure maintaining data.

The plurality of pressure holding times under the set temperature and set pressure conditions in the effective pressure holding data are sequentially compared with the pressure holding time threshold in the two-dimensional threshold array (array formed by corresponding temperature and pressure) in the initial model. And updating the minimum pressure holding time to the pressure holding time threshold of the corresponding two-dimensional threshold array under the set pressure and the set temperature, and updating the attribute value of the corresponding two-dimensional attribute array to be true. And according to the training mode, comparing the pressure holding time under all the rest temperature and pressure conditions in the effective pressure holding data with that in the initial model, completing model training and generating a leakage model.

The embodiment provides a method for predicting pressure leakage of a main air braking path, which includes the steps of training to obtain a leakage model based on relevant parameters of the main air braking path of a subway train, calculating to obtain effective data of the pressure of the main air path based on running state data of the subway train, and inputting the effective data of the pressure of the main air path into the leakage model to predict the air pressure leakage condition of the train. According to the method, effective data of the pressure of the main air path is obtained through calculation under the condition that the accuracy of the sensor is limited; the reliability and effectiveness of pressure leakage prediction data are improved, and the efficiency and accuracy of main air path pressure prediction of the subway train are enhanced; the maintenance work of the main air path of the subway train is guided in time, and the maintenance cost is reduced.

Example 2

As shown in fig. 5, the present embodiment provides a system for predicting a pressure leakage of a main brake air path, including: an acquisition module 210, a pre-processing module 220, a screening module 230, a calculation module 240, and a prediction module 250.

The acquiring module 210 is configured to acquire running state data of a target subway train in a target time period; the operating state data includes parking confirmation parameters.

A preprocessing module 220, configured to preprocess the running state data; the preprocessing comprises the steps of carrying out integrity detection and abnormal data correction on the running state data; and generates preprocessing judgment information.

And the screening module 230 is configured to screen the parking state data corresponding to the longest parking time period of the target subway train from the running state data according to the parking confirmation parameters.

And the calculating module 240 is used for calculating and obtaining the effective pressure data of the main air path in the longest parking time period based on the parking state data.

And the prediction module 250 is used for inputting the effective data of the main air path pressure into the leakage model and outputting the leakage state of the main air path pressure of the braking of the target subway train in the target time period. The leakage model is a mathematical model constructed by a two-dimensional threshold value array and a two-dimensional attribute array, the dimensionality of the two-dimensional threshold value array comprises temperature and pressure, each threshold value in the two-dimensional threshold value array represents a retention time threshold value of the pressure of a main air path under set temperature and set pressure, and the attribute value of the two-dimensional attribute array comprises a true value and a false value.

The obtaining module 210 receives the train data related to train operation collected by electronic devices such as sensors on the target subway train at regular time. The obtaining module 210 obtains the operation state data related to the main wind path pressure in the target time interval (from 12.1 early morning in 1.12.2021 to 12.2.24. 2021) from the driving data. The operation state data can comprise the current time, the pressure value of the main air path, the ambient temperature value outside the automobile, parking confirmation parameters and the like.

The preprocessing module 220 first checks the integrity of the running state data, and if the data is complete, further detects whether the data format meets a preset requirement (for example, a pkl or csv file), and if the data format meets the requirement, finally determines whether the data size meets the limit of the minimum data size. If any one of the above check results fails, a prompt message of data acquisition failure can be returned. And identifying abnormal data points corresponding to the main wind path pressure in the operation state data, such as data points of single mutation and data points exceeding a preset range (for example, the preset range is 7.5 bar-9.0 bar). The preprocessing module 220 corrects the abnormal data points by neighbor replacement, mean correction, default correction, and the like. If the correction fails, prompt information of the failure of the correction can be returned.

And selecting a plurality of parking time periods from the running state data according to the parking confirmation parameters, and selecting the longest parking time period from the plurality of parking time periods. The further screening module 230 screens the parking status data corresponding to the longest parking time period from the operating status data.

According to the pressure value corresponding to the non-repeated main wind gas circuit pressure appearing in the parking state data, the maximum pressure value and the minimum pressure value are removed, and finally the calculation module 240 screens out invalid data such as the main wind gas circuit pressure corresponding to the abnormal sampling moment to generate a pressure holding value set.

Inputting the effective data of the pressure of the main air path into a pre-trained leakage model, and outputting a leakage state result of the pressure of the target subway train in a target time period by the prediction module 250. It is to be appreciated that the prediction module 250 may also output a data state result of the main wind path pressure valid data. For example, the data state results may include data integrity analysis, preprocessing analysis, parking time presence analysis, timeout point presence analysis, pressure hold duration analysis, and the like.

In one implementation, the parking confirmation parameters include train speed, cab activation factor and air compressor operating factor; the screening module 230 includes:

the first intercepting unit 231 is configured to intercept at least one alternative parking time period from the target time period according to the train speed, the cab activation coefficient, and the air compressor work coefficient.

The first screening unit 234 is configured to screen out the longest time period from the candidate parking time periods as the longest parking time period.

And a third intercepting unit 235, configured to intercept parking state data corresponding to the longest parking time period from the running state data.

The first interception unit 231 takes the time period in which the train speed is zero, the activation coefficients of the front cab and the rear cab are zero, and the working coefficient of the air compressor is zero in each time period in the running state data as the alternative parking time period of the target subway train.

In another implementable scenario, as shown in fig. 5, the operation state data further includes a time parameter and a reference parameter, the reference parameter includes a main braking air path pressure; the parking confirmation parameters comprise the speed of the train, the activation coefficient of the cab and the working coefficient of the air compressor; the screening module 230 includes:

and a second intercepting unit 232, configured to intercept at least one primarily selected parking time period from the target time period according to the train speed, the cab activation coefficient, and the air compressor working coefficient.

And the first comparison unit 233 is configured to compare every two adjacent air pressure change values in the main braking air path pressure with the first threshold, and determine at least one alternative parking time period from the primary parking time period according to a comparison result.

The second intercepting unit 232 takes the time periods in which the train speed at each time period in the running state data is zero, the activation coefficients of the front cab and the rear cab are zero, and the working coefficient of the air compressor is zero as the initial selection parking time period of the target subway train.

And carrying out preset data processing on the pressure of the main braking air path corresponding to the plurality of initially selected parking time periods. Specifically, after removing the invalid primary parking time period in which the change difference between the adjacent air pressures in the main braking air path is greater than the first threshold (0.1bar), the first comparison unit 233 selects the remaining primary parking time period as the candidate parking time period, screens out the longest parking time period from the candidate parking time periods, and further obtains the parking state data corresponding to the longest parking time period. If all the initially selected parking time periods do not meet the conditions, prompt information of failure in obtaining the parking state data can be returned.

In one implementation, the calculation module 240 includes:

the extracting unit 241 is configured to extract a plurality of different air pressure values in the parking state data, delete the maximum and minimum air pressure values, and generate an initial air pressure array.

And a removing unit 242, configured to remove, from the initial air pressure array, an air pressure value corresponding to a sampling time interval greater than a second threshold, and generate a pressure holding set and timeout determination information.

And a generating unit 243, configured to generate the candidate pressure holding sample set and the pressure holding preset judgment information according to the pressure holding set and the parking state data.

A second comparing unit 244, configured to compare the number of samples in the candidate pressure maintenance sample set with the preset number.

And the first processing unit 245 is configured to, if the number of samples is not greater than the preset number, directly use the alternative pressure holding sample set as an effective pressure holding sample set, and generate alternative pressure holding sample judgment information and main air path pressure effective data.

And the second processing unit 246 is configured to sort the samples in the alternative pressure maintaining set according to the order from long pressure maintaining time to short pressure maintaining time if the number of the samples is greater than the preset number, screen the effective pressure maintaining sample set meeting the preset number, and generate alternative pressure maintaining sample judgment information and main air path pressure effective data.

The extraction unit 241 counts the air pressure values of all the non-repetitive main air path pressures in the parking state data. The nonrepeating air pressure values were [8.6bar, 8.7bar, 8.8bar, 8.9bar ], the maximum value of 8.9bar and the minimum value of 8.6bar among the nonrepeating air pressure values were removed, and the initial air pressure array was generated as [8.7bar, 8.8bar ].

The culling unit 242 culls two air pressure values with sampling time intervals larger than a second threshold (for example, 10min) from the initial air pressure array, and generates a pressure hold set of [8.7bar, 8.8bar ], and timeout information. The timeout information may include whether there is an abnormal data point using the time interval in the initial air pressure array, that is, an air pressure point of the timeout data.

Comparing the number of samples in the alternative pressure maintaining set samples with the preset number N, determining an effective pressure maintaining sample set according to the comparison result, and generating effective data of the pressure of the main air path and judgment information of the alternative pressure maintaining samples by the first processing unit 245 or the second processing unit 246 according to the effective pressure maintaining sample set. The candidate pressure-holding sample judgment information includes information on whether the number of samples is too large, too small, or reasonable.

In an implementable scenario, the generating unit 243 is specifically configured to:

at least one target pressure maintenance value is selected from the set of pressure maintenance values.

And after comparing the difference value between the air pressure value and the target pressure holding value in the parking state data with a third threshold value, intercepting a plurality of candidate target pressure holding time periods from the longest parking time period according to the comparison result.

And screening the longest target pressure maintaining time period from the alternative target pressure time periods, and judging whether the longest target pressure maintaining time period meets a preset condition.

And if the target pressure holding state data corresponding to the longest target pressure holding time period intercepted from the parking state data is met, generating an alternative pressure holding sample set.

The generating unit 243 may select one target pressure maintenance value (e.g., 8.7bar) at a time from the pressure maintenance set, and screen out a plurality of candidate target pressure maintenance values that are greater than or equal to 8.7bar and do not exceed one sensitivity unit (a third threshold, e.g., 0.1bar) from the parked state data in step S432.

The generating unit 243 further screens out the longest target pressure holding time period in each candidate target pressure holding time period, and then determines whether the sampling time corresponding to the longest target pressure holding time period starts at the starting parking time of the target subway train or ends at the ending parking time of the target subway train. If so, rejecting the longest target pressure holding time period, if not, continuously judging whether the length of the pressure holding time period corresponding to the longest target pressure holding time period is less than 1 minute, if so, rejecting the longest target pressure holding time period, and if not, retaining the longest target pressure holding time period.

In one embodiment, the prediction system is trained to derive the leak model by:

the system comprises an acquisition module, a storage module and a control module, wherein the acquisition module is used for acquiring pressure maintenance data of a plurality of normal main air circuit pressures of the subway train in a historical time period.

And the training module is used for inputting the pressure maintaining data into the initial model for training so as to obtain a leakage model.

Specifically, the acquisition module acquires pressure maintenance data of the pressure of the main wind gas circuit of a certain subway train in a healthy state (without leakage or with small reasonable leakage) during 12 and 21 days in 2018 and 21 days in 2019 and 2 and 28 days in 2019 and 6 and 29 days in 2019 and 7 and 9 days in 2019. And inputting the pressure maintenance data of the two time periods into an initial model for training through a training module in a day unit. It will be appreciated that the initial state threshold arrays for this initial model are all set to 0, and the attribute arrays are all set to false.

The pressure maintenance data includes standard parameters, and the training module includes:

and a selection unit for selecting the pressure maintenance data having the standard parameter within the set range as effective pressure maintenance data.

And the correcting unit is used for correcting the pressure holding time threshold of the two-dimensional threshold array in the initial model based on the pressure holding time in the effective pressure holding data to generate a leakage model.

Specifically, under the condition that the standard parameters can include temperature and pressure, the temperature change range is preset to be-10-50 ℃, the interval is 1 ℃, the pressure change range is set to be 7.5-9bar, and the interval is 0.1 bar; the selection unit takes the data in the pressure holding data which simultaneously meets the above standard parameter setting conditions as effective pressure holding data.

The plurality of pressure holding times under the set temperature and set pressure conditions in the effective pressure holding data are sequentially compared with the pressure holding time threshold values in the two-dimensional threshold value array (array constituted by the corresponding temperature and pressure) in the initial model (as shown in fig. 6). The correction unit updates the minimum pressure holding time to the pressure holding time threshold of the corresponding two-dimensional threshold array at the set pressure and the set temperature, and updates the attribute value of the corresponding two-dimensional attribute array to true. According to the training mode, the pressure holding time under all the rest temperature and pressure conditions in the effective pressure holding data is compared with that in the initial model, so that the model training is completed (as shown in fig. 7), and the leakage model is generated.

The embodiment provides a prediction system for pressure leakage of a main air braking gas circuit, a leakage model is obtained through training based on relevant parameters of the main air braking gas circuit of a subway train, effective data of the pressure of the main air gas circuit is obtained through calculation by a calculation module based on running state data of the subway train, and the effective data of the pressure of the main air gas circuit is input into the leakage model by a prediction module to predict the air pressure leakage condition of the train. According to the method, effective data of the pressure of the main air path is obtained through calculation under the condition that the accuracy of the sensor is limited; the reliability and effectiveness of pressure leakage prediction data are improved, and the efficiency and accuracy of main air path pressure prediction of the subway train are enhanced; the maintenance work of the main air path of the subway train is guided in time, and the maintenance cost is reduced.

Example 3

Fig. 8 is a schematic structural diagram of an electronic device provided in this embodiment. The electronic device includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the program to implement the method for predicting the main braking air path pressure leakage according to embodiment 1, and the electronic device 60 shown in fig. 8 is only an example and should not bring any limitation to the function and the scope of the embodiment of the present invention.

The electronic device 60 may be embodied in the form of a general purpose computing device, which may be a server device, for example. The components of the electronic device 60 may include, but are not limited to: the at least one processor 61, the at least one memory 62, and a bus 63 connecting the various system components (including the memory 62 and the processor 61).

The bus 63 includes a data bus, an address bus, and a control bus.

The memory 62 may include volatile memory, such as Random Access Memory (RAM)621 and/or cache memory 622, and may further include Read Only Memory (ROM) 623.

The memory 62 may also include a program/utility 625 having a set (at least one) of program modules 624, such program modules 624 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.

The processor 61 executes various functional applications and data processing, such as a method for predicting a main brake air path pressure leak according to embodiment 1 of the present invention, by executing a computer program stored in the memory 62.

The electronic device 60 may also communicate with one or more external devices 64 (e.g., keyboard, pointing device, etc.). Such communication may be through an input/output (I/O) interface 65. Also, model-generating device 630 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN) and/or a public network, such as the Internet) via network adapter 66. As shown, network adapter 66 communicates with the other modules of model-generating device 60 via bus 63. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the model-generating device 60, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, etc.

It should be noted that although in the above detailed description several units/modules or sub-units/modules of the electronic device are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functionality of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.

Example 4

The present embodiment provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method of embodiment 1 for predicting a main brake wind path pressure leak.

More specific examples, among others, that the readable storage medium may employ may include, but are not limited to: a portable disk, a hard disk, random access memory, read only memory, erasable programmable read only memory, optical storage device, magnetic storage device, or any suitable combination of the foregoing.

In a possible implementation, the invention may also be implemented in the form of a program product comprising program code for causing a terminal device to perform the steps of implementing the method of predicting a main brake air path pressure leak of embodiment 1, when said program product is run on the terminal device.

Where program code for carrying out the invention is written in any combination of one or more programming languages, the program code may be executed entirely on the user device, partly on the user device, as a stand-alone software package, partly on the user device and partly on a remote device or entirely on the remote device.

While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications are within the scope of the invention.

Claims

1. A prediction method for pressure leakage of a main brake air path is characterized by comprising the following steps:

2. The method of predicting main brake air path pressure leak of claim 1, wherein the parking validation parameters include train speed, cab activation coefficient, and air compressor operating coefficient;

or the like, or, alternatively,

3. The method of predicting main brake air path pressure leak of claim 1, wherein said step of calculating available main air path pressure data for the longest parking time period based on the parking status data comprises:

4. The method of predicting a main brake air circuit pressure leak of claim 3, wherein the step of generating a set of alternative pressure maintenance samples from the set of pressure maintenance and the parked state data comprises:

5. The method of predicting main brake air path pressure leakage according to claim 1, wherein the prediction method is trained to obtain the leakage model by the steps of:

6. The method of predicting a main brake air path pressure leak of claim 5, wherein the pressure maintenance data includes standard parameters, and the step of inputting the pressure maintenance data into an initial model for training to obtain the leak model includes:

7. The method for predicting main brake air circuit pressure leakage according to claim 4, wherein after the operation state data of the target subway train in the target time period is obtained, the method further comprises the following steps:

8. A system for predicting brake main duct pressure leakage, the system comprising:

9. The system of claim 8, wherein the parking validation parameters include train speed, cab activation coefficient, and air compressor operating coefficient; the screening module includes:

or the like, or, alternatively,

10. The system of claim 8, wherein the calculation module comprises:

11. The system for predicting main brake air path pressure leak of claim 10, wherein the generating unit is specifically configured to:

12. The system of claim 8, wherein the prediction system is trained to derive the leak model by:

13. The system of claim 12, wherein the pressure maintenance data includes standard parameters, the training module including:

14. The system for predicting a main brake air path pressure leak of claim 10, further comprising:

15. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the method for predicting a main brake air path pressure leak of any of claims 1-7 when executing the computer program.

16. A computer-readable storage medium, having stored therein a computer program which, when executed by a processor, implements the method of predicting main brake wind path pressure leak of any of claims 1-7.