阅读说明：本技术 一种基于tcms数据的列车制动系统管路泄漏诊断方法 (Train braking system pipeline leakage diagnosis method based on TCMS data ) 是由 陈美霞 梁师嵩 胡佳乔 蒋红果 吴强 滑瑾 蒋陵郡 倪弘韬 李�瑞 于 2021-08-13 设计创作，主要内容包括：本发明提供一种基于TCMS数据的列车制动系统管路泄漏诊断方法,首先选择空压机工作率、制动频率和制动缸保压率三项指标参数作为三维重构特征,建立孤立森林模型进行异常点的检测,选择连续周期的有效值对应的TCMS数据进行线性回归模型拟合,根据得到的线性回归模型的参数判断列车制动系统管路是否存在泄漏故障。本发明既节省了人力成本,也不需要增添设备成本,同时本发明具有一定噪声容错性。(The invention provides a train braking system pipeline leakage diagnosis method based on TCMS data. The invention saves labor cost, does not need to add equipment cost, and has certain noise fault tolerance.)

1. A train braking system pipeline leakage diagnosis method based on TCMS data is characterized by comprising the following steps:

step 1: the method comprises the steps of periodically acquiring TCMS data of a train, initializing and carrying out anti-interference processing on the TCMS data, and calculating the braking frequency of a train brake cylinder, the working rate of an air compressor and the pressure maintaining rate of the brake cylinder in the current period based on the processed TCMS data; establishing a three-dimensional coordinate system by taking the braking frequency of a brake cylinder as a y axis, the working rate of an air compressor as a z axis and the pressure maintaining rate of the brake cylinder as an x axis; drawing the braking frequency of the train brake cylinder in the current period, the working rate of the air compressor and the pressure maintaining rate of the brake cylinder into a three-dimensional coordinate system to be used as three-dimensional characteristic points of the train braking system pipeline in the current period; the TCMS data comprises: total jump number sigma N of brake states of one random brake cylinder in train in sampling period_brakeTotal time of operation T of the braking system of the train in the sampling period_allAnd total working time sigma T of all air compressors in the train in the sampling period_fillTotal time of pressure in all brake cylinders in a train during a sampling period

Step 2: judging whether the three-dimensional feature points of the current period are abnormal feature points or not according to the isolated forest model, if not, taking the feature points as effective values of the current period, and turning to the step 3; otherwise, deleting the three-dimensional characteristic points of the current period, taking the effective value of the previous period as the effective value of the current period, and turning to the step 3;

and step 3: forming a group of current TCMS data groups by the effective values of the current period and TCMS data corresponding to the effective values of the first N periods of the current period; and constructing a linear regression model as follows:

wherein A, B and C are parameters, and Const is 1;

fitting the linear regression model by using the current TCMS data group and preset historical data, and if the B value obtained after the current TCMS data group fits the linear regression model exceeds a preset B value range and the B values obtained after the previous Q groups of TCMS data groups of the current TCMS data group fit the linear regression model exceed the preset B value range, determining that the brake pipeline of the train has a leakage fault;

and if the C value obtained after the linear regression model is fitted to the current TCMS data group exceeds the preset C value range, and the C values obtained after the linear regression model is fitted to the front Q TCMS data group of the current TCMS data group also exceed the preset C value range, determining that the leakage fault exists in the total air pipeline of the train.

2. The method for diagnosing the pipeline leakage of the train brake system based on the TCMS data as claimed in claim 1, wherein the expression of the braking frequency of the train brake cylinder is as follows:

the expression of the working rate of the air compressor is as follows:

the expression of the brake cylinder packing pressure ratio is as follows:

3. the method for diagnosing the leakage of the train braking system pipeline based on the TCMS data as claimed in claim 2, wherein the preset B value range and the preset C value range obtained in the step 3 are specifically as follows: acquiring TCMS data under the healthy state of a train braking system pipeline, initializing and carrying out anti-interference processing on the TCMS data, obtaining a plurality of three-dimensional characteristic points according to the processed TCMS data, drawing the three-dimensional characteristic points into a linear plane under the three-dimensional coordinate system in the step 1, and fitting the linear plane Z to A + x + B + y + C; x represents the x-axis and y represents the y-axis, such that the range of values of B is preset based on the value of B in the linear plane and the range of values of C is preset based on the value of C in the linear plane.

4. The method of claim 1, wherein the sampling period has a cycle length of 1 day.

Technical Field

The application belongs to the technical field of brake system maintenance.

Background

The train braking system is one of important components of a train, and once the train braking system breaks down, serious railway accidents such as rail break-in and rail break-back are possibly caused; the braking system has many pipelines and pipe joints, so that potential leakage faults are easy to occur, and after the leakage faults occur, the performance of the braking system is reduced, and the train operation is influenced. Accordingly, train brake system piping needs to be serviced to discover the presence or potential of a brake system piping leak. At present, methods such as manual overhaul or pressure maintaining test and the like are mostly adopted for diagnosing the pipeline leakage of the brake system, the manual overhaul of the pipeline leakage of the brake system is mainly included in daily inspection projects, the methods such as ear-to-ear inspection, hand touch, soapy water and the like are used, the method is original, the efficiency is low, the consumed labor cost is high, and in addition, overhaul personnel have certain experience, and the possible omission problem of manual overhaul cannot be eliminated; the pressure maintaining test judges whether leakage occurs or not through the pressure drop rate of the train under the static working condition, and positions the section of the pipeline where the leakage occurs through opening and closing of the cock. Although the test method can detect the leakage more accurately, the operation time of the urban rail transit train is long, the time of returning to the warehouse and the human resource are required for carrying out the test, and even the application schedule of the train is influenced. The patent application document CN112141069A adopts a predictive maintenance technology to perform overall evaluation on the performance of the brake system by additionally configuring a health management host edge device for the train and additionally installing an additional sensor. However, the scheme needs additional equipment, so that the cost is increased, and the additional equipment is not suitable for configuration on an operated line due to a long modification period. And the implementation mode of the performance early warning module of the air supply system in the file is as follows: and recording the total wind pressure and time before the power failure of the motor train unit, judging after electrifying again, recording the reduction rate of the total wind pressure if the power failure time is less than the total wind pressure complete leakage time of the train, and giving early warning on the leakage condition of the whole train pipeline. The early warning scheme needs to power off the train, and the train operation can be influenced.

Disclosure of Invention

The purpose of the invention is as follows: in order to solve the problems in the prior art, the invention provides a train brake system pipeline leakage diagnosis method based on TCMS data.

The technical scheme is as follows: the invention provides a train braking system pipeline leakage diagnosis method based on TCMS data, which specifically comprises the following steps:

step 1: the method comprises the steps of periodically acquiring TCMS data of a train, initializing and carrying out anti-interference processing on the TCMS data, and calculating the braking frequency of a train brake cylinder, the working rate of an air compressor and the pressure maintaining rate of the brake cylinder in the current period based on the processed TCMS data; establishing a three-dimensional coordinate system by taking the braking frequency of a brake cylinder as a y axis, the working rate of an air compressor as a z axis and the pressure maintaining rate of the brake cylinder as an x axis; drawing the braking frequency of the train brake cylinder in the current period, the working rate of the air compressor and the pressure maintaining rate of the brake cylinder into a three-dimensional coordinate system to be used as three-dimensional characteristic points of the train braking system pipeline in the current period; the TCMS data comprises: total jump number sigma N of brake states of one random brake cylinder in train in sampling period_brakeTotal time of operation T of the braking system of the train in the sampling period_allAnd total working time sigma T of all air compressors in the train in the sampling period_fillTotal time of pressure in all brake cylinders in a train during a sampling period

Step 2: judging whether the three-dimensional feature points of the current period are abnormal feature points or not according to the isolated forest model, if not, taking the feature points as effective values of the current period, and turning to the step 3; otherwise, deleting the three-dimensional characteristic points of the current period, taking the effective value of the previous period as the effective value of the current period, and turning to the step 3;

and step 3: forming a group of current TCMS data groups by TCMS data corresponding to effective values of the current period and the first N periods of the current period; and constructing a linear regression model as follows:

wherein A, B and C are parameters, and Const is 1;

fitting the linear regression model by using the current TCMS data group and preset historical data, and if the B value obtained after the current TCMS data group fits the linear regression model exceeds a preset B value range and the B values obtained after the previous Q groups of TCMS data groups of the current TCMS data group fit the linear regression model exceed the preset B value range, determining that the brake pipeline of the train has a leakage fault;

and if the C value obtained after the linear regression model is fitted to the current TCMS data group exceeds the preset C value range, and the C values obtained after the linear regression model is fitted to the front Q TCMS data group of the current TCMS data group also exceed the preset C value range, determining that the leakage fault exists in the total air pipeline of the train.

Further, the expression of the braking frequency of the train brake cylinder is as follows:

the expression of the working rate of the air compressor is as follows:

the expression of the brake cylinder packing pressure ratio is as follows:

further, the preset B value range and the preset C value range obtained in the step 3 are specifically: acquiring TCMS data under the healthy state of a train braking system pipeline, initializing and carrying out anti-interference processing on the TCMS data, obtaining a plurality of three-dimensional characteristic points according to the processed TCMS data, drawing the three-dimensional characteristic points into a linear plane under the three-dimensional coordinate system in the step 1, and fitting the linear plane Z to A + x + B + y + C; x represents the x-axis and y represents the y-axis, such that the range of values of B is preset based on the value of B in the linear plane and the range of values of C is preset based on the value of C in the linear plane.

Further, the cycle length of the sampling period is 1 day.

Has the advantages that:

1. the pipeline leakage detection is periodically carried out based on the designed data model, so that the labor cost is saved; in addition, by a data driving method, leakage detection is carried out by accessing train TCMS data on the basis of not additionally arranging a sensor, and equipment cost is not increased; meanwhile, the method has certain noise tolerance, because the running states of the trains are inconsistent every day, the trains are likely to run on the line one day and are in the warehouse one day, so that the actual air consumption amount of each day cannot be compared independently, and a mapping relation among a plurality of physical quantities needs to be established. Therefore, the working rate of the air compressor, the braking frequency and the pressure maintaining rate of the brake cylinder are selected as model characteristics for modeling. TCMS data is transmitted by 4G vehicles, transmission loss is caused by the quality of 4G signals in actual data, large data loss occurs, the lengths of data which can be acquired by each train every day are inconsistent, and therefore the influence of the absolute length of the data can be eliminated by adopting a rate related concept as a model characteristic.

2. The invention performs mixed modeling on the two types of leakage, namely 'total braking air pipeline leakage' and 'brake cylinder and connected pipeline leakage', and uses different parameters of the model to represent the leakage phenomena at different positions, thereby avoiding the problem that the two types of leakage can not be considered and modeled respectively due to the stronger coupling phenomenon existing in the mechanism.

Drawings

FIG. 1 is a schematic signal flow diagram of a wind supply and braking system;

FIG. 2 is a schematic diagram of the logic structure of the method of the present invention;

FIG. 3 is a simplified model diagram of the air supply and braking system provided by the present invention;

FIG. 4 is a schematic diagram of a feature point distribution according to the present invention;

FIG. 5 is a schematic diagram of a feature point distribution and a fitting plane provided by the present invention;

FIG. 6 is a schematic diagram of data results of an actual vehicle according to the present invention.

Detailed Description

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate an embodiment of the invention and, together with the description, serve to explain the invention and not to limit the invention.

For an urban rail train, the signal flow directions of an air supply and brake system are shown in fig. 1, in the brake system, a main air pipe is a pipeline which is connected with a main air cylinder, the brake system and other air utilization equipment, the main air cylinder is filled with air by an air source system to obtain pressure air, the pressure air is stored, and all air utilization modules on the train are supplied with the air, and the urban rail train air supply and brake system air supply system comprises: the air for a brake system, the air for an air spring, the air for a pneumatic plug and pull door, the waste exhaust of an air conditioner, the pantograph lifting, the air cooling of high-temperature elements and the like.

The leakage fault of the main air pipeline may be caused by that the leakage of pressure air is generated in the long-time service process due to the limited sealing performance of the pipeline, the rubber element and the pipeline joint, and the final result of the leakage fault can lead to the acceleration of the pressure reduction rate of the main air cylinder, thereby leading to the frequent start and stop of the air compressor.

The leakage fault of the brake cylinder and the connected pipeline can be caused by that the pressure maintaining performance of the valve and the sealing performance of a pipeline joint are reduced in the pressure maintaining process of the brake cylinder, but because the pressure of the brake cylinder is used as a controlled quantity, if the pressure is reduced due to leakage in the actual operation process, the BCU can automatically perform pressure control to supplement air, and the air supplementing process is not represented by a corresponding instruction signal in a TCMS signal, so that effective leakage characteristics cannot be extracted from the pressure signal of the brake cylinder. Therefore, the leakage phenomenon of the brake pipeline is subjected to mechanism analysis and traced to an air supply system, and the leakage of the brake pipeline can directly lead to the increase of the air consumption of the whole train, so that the starting and stopping of the air compressor are more frequent.

Based on the above thought, as shown in fig. 2, this embodiment provides a train brake system pipeline leakage diagnosis method based on TCMS data, where the TCMS data first enters the inside of the model through the initialization and anti-interference module, and through a four-layer structure of data processing, state monitoring, health assessment and fault early warning, characteristic parameters of the working state of the air supply/brake system are evaluated by counting three items, namely the working rate of the air compressor, the brake frequency and the pressure holding rate of the brake cylinder, and a total air pipeline leakage item, the brake cylinder and a connected pipeline leakage item are decoupled, so as to complete diagnosis and differentiation of leakage of the above two types of pipelines. The embodiment specifically includes:

1. mechanism model simplification

According to the mechanism analysis of the air supply and brake system, no matter the total air pipeline leaks or the brake cylinder and the connected pipelines leak, the working time of the air compressor is increased, and the characteristics of the total air pipeline leakage and the brake pipeline leakage need to be represented respectively through different characteristics, so that the leaking position is located. Therefore, the model is established by analyzing the mechanism relation between the starting and stopping of the air compressor, the pressure air consumed in the braking process and the pressure air leaked from the air compressor and the air compressor.

In summary, the entire braking system and the air supply system are abstracted into a topology as shown in fig. 3

2. Selecting relevant features

Firstly, constructing model characteristics, and selecting the following three index parameters as three-dimensional reconstruction characteristics through understanding service and data:

the working rate of the air compressor, the braking frequency and the pressure maintaining rate of the brake cylinder.

The mapping relation between the air consumption and the leakage phenomenon of the whole vehicle can be reflected. And can distinguish whether the root cause of the increase of the air consumption of the whole automobile is the leakage of a main air pipeline or the leakage of a brake cylinder and a connecting pipeline.

The three-dimensional reconstruction characteristic calculation method comprises the following steps:

1) the working rate of the air compressor: total working time Σ T of all air compressors (two in this embodiment) in the train_fillAnd the total working time T of the brake system_allA ratio of (A) to (B);

2) braking frequency of the brake cylinder: selecting one brake cylinder optionally, and calculating the number sigma N of jumping of the brake state of the brake cylinder_brakeAnd the total working time T of the brake system_allA ratio of (A) to (B);

3) the brake cylinder pressure maintaining rate is as follows: total time of pressure in any brake cylinder and total working time T of brake system_allThe ratio of (A) to (B):

b_stat1 indicates that there is pressure in the brake cylinder.

And establishing a three-dimensional coordinate system by taking the braking frequency of the brake cylinder as a y axis, the working rate of the air compressor as a z axis and the pressure maintaining rate (namely the braking state duration ratio) of the brake cylinder as an x axis.

Since the leakage belongs to a slightly gradual fault mode, the characteristics are possibly buried in the problems caused by data noise or data quality when the judgment is carried out in a short period, the characteristics cannot be effectively represented on the data, and the daily working state of the train is not determined, so that the data statistical period is put to one day, namely, the daily data of each train outputs a point (characteristic point) in a three-dimensional space, and the range of a health domain is determined through the calibration of historical data. The distribution of the feature points in the three-dimensional space is shown in fig. 4.

3. Leak detection based on calculated characteristics

1) An abnormality detection stage:

firstly, establishing an isolated forest model for anomaly detection, and judging whether a three-dimensional characteristic point of a certain day required to be diagnosed exceeds a normal data boundary due to transmission anomaly, working condition anomaly and a debugging state of a train. If the point does not exceed the boundary, the point is a valid value, and the valid value is reserved; if the three-dimensional characteristic point exceeds the boundary, the data is discarded, and the effective value which is used for the three-dimensional characteristic point and is closest in time is used as the three-dimensional characteristic point of a certain day needing diagnosis.

2) Parameter regression calibration stage:

forming a set of current TCMS data groups from the TCMS data corresponding to the valid values of the current cycle and the first N cycles of the current cycle, where N is 10 in this embodiment

Establishing a statistical model for the air supply system and the brake system, wherein the model formula is as follows

Const represents that the ratio of the total air charging quantity to the air consumption quantity of the braking system is a fixed value in a time period, and 1 is taken out; a, B and C are three-dimensional space plane parameters which need to be calibrated by actual data, namely, a plane in the shape of z ═ A x + B x y + C is calibrated in the space and used for representing the mapping relation among the variables. The fitted plane is shown in fig. 5.

Wherein A can represent the air consumption amount which can be used for braking once per unit time of average air charging A;

b can represent the air quantity which can be leaked by a brake cylinder and a connecting pipeline in unit time in average every B unit time of air charging, and the larger the value is, the more serious the leakage phenomenon of the brake pipeline is;

c can represent the average amount of wind which can be leaked by the total wind pipeline in unit time per C unit time of wind charging, and the larger the value of the value is, the more serious the leakage phenomenon of the total wind pipeline is.

In a specific implementation, historical data and data from a recent period of time (e.g., the last 10 days) may be used to perform the fitting of the decision plane. In principle, the fitting can be performed using only the data of the last 10 days, but since there are fewer points of the fitting plane, which results in a large parameter fluctuation, the fitting needs to be performed in combination with historical data.

3) And a fault reasoning stage:

if the leakage of the main air pipeline needs to be judged, namely the time sequence change of the C value is judged, if the C value is always a larger value in a period of time, the leakage of the main air pipeline is judged.

If the leakage of the brake cylinder and the connected pipeline needs to be judged, namely the time sequence change of the B value is judged, if the B value is always a larger value in a period of time, the leakage of the brake cylinder and the connected pipeline is judged.

4. Model evaluation validation

The model evaluation uses all operation data of a certain operation subway line which is 11 trains in one month, and firstly, all data are subjected to statistical results in a three-dimensional space and a linear plane is fitted:

the parameters of the fitting plane are as follows:

Z＝13.2314*X+0.01858*Y+0.05499

13.2314 is A in the model formula and represents the air consumption amount which can be used for braking once in 13.2314 unit time on average;

0.01858 is B in the model formula, representing the air quantity which can be leaked by the brake pipeline in unit time in 0.01858 unit time of average air charging;

0.05499 is C in the model formula, representing the air quantity which can be leaked by the total air pipeline in a unit time in 0.05499 unit time of average air charging;

all parameters of the fit are reasonable in physical meaning and size relation.

And then real train time sequence data of a certain train is adopted for fitting, and a curve of the B value output by the model along with time is shown in figure 6. It can be seen that, beginning at 9/16, the model B value began to exceed the alarm threshold and remained high at both days 17 and 18, and it can be presumed that the brake cylinder and associated piping of the train had a leak failure at approximately 9/16 days.

It should be noted that the various features described in the above embodiments may be combined in any suitable manner without departing from the scope of the invention. The invention is not described in detail in order to avoid unnecessary repetition.