Intelligent train monitoring system based on Internet of Things sensor

Abstract

The invention relates to the technical field of intelligent train monitoring, in particular to an intelligent train monitoring system based on an Internet of Things sensor. According to the system, data of all parts of a train are collected and transmitted, all the data are preprocessed to obtain a preprocessed data set, the preprocessed data set is obtained, and fault diagnosis analysis is carried out according to a diagnosis model to output a monitoring result; inputting a plurality of historical operation data into the fault diagnosis model to generate an operation result, determining the diagnosis recognition rate of the current diagnosis model based on the comparison of the historical diagnosis data and the operation result, judging whether the diagnosis process of the diagnosis model is qualified or not based on the diagnosis recognition rate, and generating an instruction if the diagnosis process is unqualified, determining to carry out corresponding parameter adjustment on the system or send out an optimization notification based on the instruction; therefore, the diagnosis recognition rate is compared with the preset value to determine corresponding adjustment or send out a notification to improve the diagnosis efficiency, so that the reliability of a monitoring result generated by train fault diagnosis is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of train intelligent monitoring, and particularly relates to a train intelligent monitoring system based on an Internet of Things sensor. BACKGROUND

[0002] With the expansion of the scale of urban rail transit network, the operation density and operation time of the subway train are continuously increased, and higher requirements are put forward for the safety and reliability of the key systems such as vehicles, tracks, power supply, etc. The traditional manual inspection and fixed monitoring means have hysteresis, and it is difficult to find sudden failures (such as bearing overheating, track cracking, and abnormal overhead contact system, etc.) in time, which may lead to operation interruption and even safety accidents.

[0003] Chinese patent application publication No. CN109343425A provides an intelligent train monitoring system based on the Internet of Things, which acquires the state data of the train in real time by setting the vehicle-mounted data acquisition unit and the data transmission unit at the train end, and sends the state data to the ground monitoring detection unit in real time based on the Internet of Things protocol, so as to monitor the train according to the state data, thereby realizing real-time monitoring of the train. Although the technical scheme realizes real-time monitoring of the train through real-time acquisition of the state data, the acquired state data lacks verification, and thus the data cannot be ensured as a basis for judgment, thereby leading to low reliability of the monitoring result. SUMMARY

[0004] Therefore, the present application provides a train intelligent monitoring system based on an Internet of Things sensor, which overcomes the problem of low reliability of the monitoring result caused by the lack of data verification process and corresponding parameter adjustment of the monitoring system in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides a train intelligent monitoring system based on an Internet of Things sensor, comprising:

[0006] a data acquisition module comprising a plurality of sensors for acquiring data of each corresponding part of the train;

[0007] a data processing module connected with the data acquisition module, for receiving each data and pre-processing to obtain a pre-processed data set;

[0008] a fault diagnosis module connected with the data processing module, for acquiring the pre-processed data set and performing fault diagnosis analysis according to a diagnosis model to output a monitoring result;

[0009] a database comprising a plurality of historical operation data and a plurality of historical diagnosis data, connected with the fault diagnosis module and used to deliver a plurality of historical operation data to the fault diagnosis module to generate an operation result;

[0010] an analysis processing module connected with the fault diagnosis module and the database, configured to compare the historical diagnosis data and the running result to determine a diagnosis recognition rate of the diagnosis model;

[0011] an instruction generating module connected with the analysis processing module, configured to determine whether the diagnosis process of the diagnosis model is qualified based on the diagnosis recognition rate, and to obtain historical diagnosis recognition rates of the corresponding part based on the determination result to re-determine whether the diagnosis process is qualified, or to generate an instruction in the case of unqualification, the instruction being to re-determine a preset diagnosis recognition rate in the instruction generating module, a sampling frequency in the data collecting module, or a block quantity in the data processing module;

[0012] a control module connected with the instruction generating module, the data collecting module, and the data processing module, configured to determine whether to adjust the preset diagnosis recognition rate, the sampling frequency, or the block quantity based on the instruction.

[0013] Further, the instruction generating module is further configured to determine whether the diagnosis process of the diagnosis model is qualified based on a comparison result of the diagnosis recognition rate and the preset diagnosis recognition rate, and to calculate a diagnosis recognition variance based on the historical diagnosis recognition rates of the corresponding part and the current diagnosis recognition rate to re-determine whether the diagnosis process is qualified;

[0014] The instruction generating module is further configured to determine a reason for unqualification based on a difference between the preset diagnosis recognition rate and the diagnosis recognition rate in the case of unqualification of the diagnosis process of the diagnosis model.

[0015] Further, the instruction generating module is further configured to determine whether to reduce the preset diagnosis recognition rate based on a comparison result of the diagnosis recognition variance and a critical diagnosis recognition variance.

[0016] Further, the instruction generating module is further configured to determine to reduce the preset diagnosis recognition rate in the case that the diagnosis recognition variance is less than or equal to the critical diagnosis recognition variance, and to generate a corresponding instruction based on a comparison result of a recognition variance difference and a preset recognition variance difference to determine a reduction amplitude of the preset diagnosis recognition rate, the reduction amplitude being in a positive correlation with the recognition variance difference;

[0017] The control module is further configured to control the instruction generating module to reduce the preset diagnosis recognition rate based on the instruction.

[0018] The recognition variance difference is a difference between the critical diagnosis recognition variance and the diagnosis recognition variance.

[0019] Further, the instruction generation module is further configured to determine a reason for the diagnosis process of the current diagnosis model being unqualified based on a comparison result of the diagnosis identification difference and the critical diagnosis identification difference;

[0020] The instruction generation module is further configured to generate an instruction of issuing a notification of optimizing the diagnosis model based on the reason, or to acquire a total data amount based on the determination result and compare the total data amount with a critical total data amount to re-determine a processing;

[0021] The diagnosis identification difference is a difference between the preset diagnosis identification rate and the diagnosis identification rate, and the total data amount is a total data amount of a single data in a current monitoring period.

[0022] Further, the instruction generation module is further configured to determine to increase the sampling frequency in a case where the total data amount is less than or equal to the critical total data amount.

[0023] Further, the instruction generation module is further configured to generate a corresponding instruction to determine an increase amplitude of the sampling frequency based on a comparison result of the identification difference proportion and a preset identification difference proportion, the increase amplitude and the identification difference proportion being in a negative correlation;

[0024] The control module is further configured to control the data acquisition module to increase the sampling frequency based on the instruction;

[0025] The identification difference proportion is a ratio of the diagnosis identification difference to the critical diagnosis identification difference.

[0026] Further, the instruction generation module is further configured to determine to increase a block quantity in a preprocessing process of the data processing module in a case where the total data amount is greater than the critical total data amount.

[0027] Further, the instruction generation module is further configured to generate a corresponding instruction to determine an increase amplitude of the block quantity based on a comparison result of a total data amount difference and a preset total data amount difference, the increase amplitude and the total data amount difference being in a positive correlation;

[0028] The control module is further configured to control the data processing module to increase the block quantity based on the instruction;

[0029] The total data amount difference is a difference between the total data amount and the critical total data amount.

[0030] Further, the system further comprises an alarm module and a maintenance notification module;

[0031] The alarm module is connected with the fault diagnosis module and the instruction generation module respectively, and is configured to generate an alarm level report based on the monitoring result in a case where the diagnosis process of the diagnosis model is qualified;

[0032] The maintenance notification module is connected with the alarm module, to determine to issue a notification of a corresponding maintenance solution based on the alarm level report.

[0033] Compared with the prior art, the train intelligent monitoring system based on an Internet of Things sensor has the beneficial effects that the system acquires data of corresponding parts of the train in real time through a data acquisition module, transmits each data through a vehicle-mounted communication module, receives each data and performs preprocessing to obtain a preprocessed data set through a data processing module, acquires the preprocessed data set and performs fault diagnosis analysis according to a diagnosis model to output a monitoring result through a fault diagnosis module, inputs a plurality of historical operation data into the fault diagnosis module to generate an operation result, inputs a plurality of historical diagnosis data into an analysis processing module and acquires the operation result, and determines a diagnosis recognition rate of the current diagnosis model based on comparison of the historical diagnosis data and the operation result, determines whether the diagnosis process of the diagnosis model is qualified based on the diagnosis recognition rate, and the instruction generation module generates an instruction in the case of unqualified, the control module determines the preset diagnosis recognition rate in the analysis processing module, the sampling frequency in the data acquisition module, the number of blocks in the data processing module, or issues a notification of the diagnosis model in the optimized fault diagnosis module based on the instruction; in this way, the adjustment of the corresponding parameters is determined based on comparison of the diagnosis recognition rate and the preset value to improve the diagnosis efficiency, thereby improving the reliability of the monitoring result generated by the fault diagnosis of the train.

[0034] Further, when the diagnosis recognition rate and the preset diagnosis recognition rate are determined, the diagnosis recognition variance is further acquired based on a plurality of historical diagnosis recognition rates and the current diagnosis recognition rate, and the diagnosis process is further determined based on the diagnosis recognition variance and a critical diagnosis recognition variance, thereby improving the accuracy of the diagnosis process determination.

[0035] Further, when the diagnosis recognition variance and the critical diagnosis recognition variance are determined, the diagnosis recognition rate is greater than the preset diagnosis recognition rate when the acquired data itself has no problem, the determination criterion can be corrected by reducing the preset diagnosis recognition rate, thereby increasing the probability of the diagnosis recognition rate being greater than the preset diagnosis recognition rate, and more qualified monitoring results are output.

[0036] Further, when the preset diagnosis recognition rate is determined to be reduced, the reduction amplitude of the preset diagnosis recognition rate is determined through comparison of the recognition variance difference and the preset recognition variance value, so that the preset diagnosis recognition rate can be reduced more accurately, thereby improving the monitoring efficiency.

[0037] Further, when it is determined that the diagnosis process is unqualified, the application further determines the corresponding processing mode through the comparison of the diagnosis identification difference and the critical diagnosis identification difference, including: comparing the total data amount with the critical total data amount to re-determine the corresponding processing, or issuing a notification of optimizing the diagnosis model.

[0038] Further, when it is determined that the data acquisition process has problems through the comparison result of the total data amount and the critical total data amount, the application further determines the increasing range of the sampling frequency in the data acquisition process based on the comparison of the identification difference proportion and the preset identification difference proportion, so that the sampling frequency can be increased more accurately, thereby improving the monitoring efficiency.

[0039] Further, when it is determined that the data processing process has problems through the comparison result of the total data amount and the critical total data amount, the application further determines the increasing range of the block quantity in the data processing process based on the comparison of the total data amount difference K and the preset total data amount difference K, so that the block quantity can be increased more accurately, thereby improving the monitoring efficiency.

[0040] Further, when it is determined that the monitoring result is accurate, the application generates different alarm levels based on different fault conditions, thereby quantifying the priority of the fault and determining to issue different maintenance scheme notifications based thereon, so as to optimize the allocation of maintenance resources. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A module schematic diagram of the train intelligent monitoring system based on the Internet of Things sensor of the application;

[0042] Figure 2 A flowchart of the train intelligent monitoring method based on the Internet of Things sensor of the application;

[0043] Figure 3 A logic determination diagram for determining whether the diagnosis process is qualified and the corresponding processing based on the diagnosis identification rate of the application;

[0044] Figure 4 A logic determination diagram for determining the reason why the diagnosis process is unqualified and the corresponding processing based on the diagnosis identification difference of the application. DETAILED DESCRIPTION

[0045] In order to make the objects and advantages of the application clearer, the application will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the application, and do not limit the protection scope of the application.

[0046] The preferred embodiments of the application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the application, and are not intended to limit the protection scope of the application.

[0047] It should be noted that in the description of the present application, unless otherwise explicitly specified and limited, the term "connection" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or internal communication of two elements. For those skilled in the art, the specific meaning of the above-mentioned term in the present application can be understood according to the specific circumstances.

[0048] Please refer to Figure 1 As shown in the figure, it is a module schematic diagram of the train intelligent monitoring system based on Internet of Things sensor in the embodiment. The system includes a data acquisition module, a vehicle-mounted communication module, a data processing module, a fault diagnosis module, a database, an analysis processing module, an instruction generation module and a control module. The data acquisition module includes a plurality of sensors for collecting data of corresponding parts of the train, wherein the data acquisition mode includes real-time acquisition mode and periodic acquisition mode, and different acquisition modes are selected for data acquisition at different scenes and positions of the train; the vehicle-mounted communication module is further included, the data acquisition module and the data processing module are connected through the vehicle-mounted communication module, and the collected data are sequentially transmitted to the data processing module; the data processing module is indirectly connected with the data acquisition module through the vehicle-mounted communication module, and is used to receive the data and perform preprocessing to obtain a preprocessed data set; the fault diagnosis module is connected with the data processing module, and is used to obtain the preprocessed data set and perform fault diagnosis analysis according to a diagnosis model to output a monitoring result; the database includes a plurality of historical operation data and a plurality of historical diagnosis data, which are connected with the fault diagnosis module and used to transmit a plurality of historical operation data to the fault diagnosis module to generate an operation result; the analysis processing module is connected with the fault diagnosis module and the database respectively, and is used to obtain a plurality of historical diagnosis data and the operation result and perform comparison to determine a diagnosis recognition rate of the current diagnosis model; the instruction generation module is connected with the analysis processing module, and is used to determine whether the diagnosis process of the diagnosis model is qualified based on the diagnosis recognition rate, and based on the determination result, a plurality of historical diagnosis recognition rates are re-determined to determine whether the diagnosis process is qualified, or an instruction is generated in the case of unqualified, the instruction being a pre-set diagnosis recognition rate in the instruction generation module, or a sampling frequency in the data acquisition module, or a block quantity in the data processing module; the control module is connected with the instruction generation module, the data acquisition module and the data processing module respectively, and is used to determine adjustment of the pre-set diagnosis recognition rate, or the sampling frequency, or the block quantity based on the instruction.

[0049] Please refer to Figure 2As shown, it is a flow diagram of the train intelligent monitoring method based on the Internet of Things sensor of the present embodiment. The method comprises at least the following steps:

[0050] S1: arranging a plurality of sensors, collecting data of each corresponding part of the train through the sensors;

[0051] S2: acquiring each of the data and transmitting in turn;

[0052] S3: receiving each of the data and preprocessing to obtain a preprocessed data set;

[0053] S4: constructing a diagnostic model, inputting the preprocessed data set into the diagnostic model for fault diagnosis analysis to output a monitoring result;

[0054] S5: feeding a plurality of historical operation data into the diagnostic model to generate an operation result;

[0055] S6: acquiring the operation result and comparing with a plurality of historical diagnosis data to determine the diagnosis recognition rate of the current diagnostic model;

[0056] S7: determining whether the diagnosis process of the diagnostic model is qualified based on the diagnosis recognition rate, and based on the determination result, re-determining whether the diagnosis process is qualified based on a plurality of historical diagnosis recognition rates, or generating an instruction in the case of unqualified, the instruction being to re-determine a preset diagnosis recognition rate set, or to re-determine a sampling frequency in the data collection process, or to re-determine a block quantity in the preprocessing process;

[0057] S8: adjusting the preset diagnosis recognition rate, or adjusting the sampling frequency, or adjusting the block quantity based on the instruction.

[0058] Specifically, in the embodiment, the corresponding parts on the train include wheel bearings, traction motors, and fans, and the arranged sensors are high-precision sensor arrays, for example, MEMS accelerometers arranged on the wheelset to collect bearing vibration data, infrared temperature measurement modules to collect traction motor temperature data, voltage sensors to collect fan operating voltage, and fiber Bragg grating sensors to monitor the stress of the car body structure to collect strain data; the deployment of the sensors uses a train-track collaborative architecture, with both on-board sensors and trackside sensors; a vehicle communication module can be provided on the train to receive the collected data in real time. The priority of each data can be determined based on a machine learning algorithm, and the data is transmitted in order according to the priority to ensure that data with relatively high real-time performance is transmitted first, avoiding being blocked by non-critical data. For example, the data with the highest priority is determined to be the most critical data, which is transmitted first using 5G, and data with low priority can be batched to save bandwidth; the preprocessing process includes cleaning, filtering, and blocking the data. The monitoring results include the fault type and corresponding fault parameters at each corresponding part; by feeding several historical operation data into the fault diagnosis module, the diagnosis model analyzes the historical operation data to obtain a diagnosis result, and the diagnosis result and several historical diagnosis data are input into the analysis processing module, and then the diagnosis result and historical diagnosis data are analyzed and compared to determine the diagnosis recognition rate of the current diagnosis model. Comparing the diagnosis recognition rate with the preset value can determine whether the diagnosis process of the current diagnosis model is qualified; the diagnosis recognition rate refers to the probability of correctly identifying abnormal state data in the fault diagnosis process, which is used to determine whether the monitoring result obtained by diagnosing a part is reliable. For example, taking bearing crack fault recognition as an example, if the actual diagnosis recognition rate is 96% and the preset value is 95%, it is determined that the monitoring result for the current bearing crack is accurate and reliable, and the current monitoring result can be used for subsequent use. In the embodiment, the values of the corresponding preset parameters or critical parameters are determined by analyzing historical data obtained in previous train monitoring and statistical methods, or some values can also be set according to industry standards.

[0059] Please refer to Figure 3As shown, it is a logical determination diagram for determining whether the diagnosis process is qualified and corresponding processing based on the diagnosis recognition rate of the present embodiment. The instruction generation module is also used to determine whether the diagnosis process of the diagnosis model is qualified based on the comparison result of the diagnosis recognition rate and the preset diagnosis recognition rate, and to obtain the diagnosis recognition variance calculated based on the current diagnosis recognition rate and a plurality of historical diagnosis recognition rates of the corresponding part based on the determination result; the instruction generation module is also used to determine the unqualified reason based on the difference between the preset diagnosis recognition rate and the diagnosis recognition rate when it is determined that the diagnosis process of the diagnosis model is unqualified.

[0060] Specifically, in the present embodiment, the preset diagnosis recognition rate V0 can be divided into a first preset diagnosis recognition rate V1 and a second preset diagnosis recognition rate V2, and V1 and V2 are compared with the diagnosis recognition rate V to refine the determination process, V1 ∈ [95%, 98%], V2 ∈ [90%, 93%]; the process of comparing V with V1 and V2 is as follows:

[0061] If V is greater than or equal to V1, it means that the current V is relatively large and meets the abnormal diagnosis requirement, and it can be determined that the diagnosis process of the current diagnosis model is qualified. If V is less than V1 and greater than or equal to V2, a plurality of historical diagnosis recognition rates and the current diagnosis recognition rate of the corresponding part can be obtained, and variance calculation is performed based thereon to obtain a diagnosis recognition variance F, and then the diagnosis recognition variance F is compared with a critical diagnosis recognition variance F0 to further determine whether the current diagnosis process is qualified, thereby outputting more qualified monitoring results and improving the diagnosis efficiency. At this time, F is compared with F0 based on the determination result to avoid relying on a single index and improve the accuracy of the determination result. If V is less than V2, it means that the current V is relatively small and cannot meet the abnormal diagnosis requirement, and it can be determined that the diagnosis process of the current diagnosis model is unqualified, and the reason can be determined by the difference between V0 and V at this time, and more specifically, the difference between V2 and V, V1 and V2 in the comparison process are specific numerical values.

[0062] Further, the instruction generation module is also used to determine whether to reduce the preset diagnosis recognition rate based on the comparison result of the diagnosis recognition variance and the critical diagnosis recognition variance.

[0063] Specifically, in the present embodiment, the critical diagnosis recognition variance F0 ∈ [0.3, 0.35] is set; the comparison process of the diagnosis recognition variance F and F0 is as follows:

[0064] If F is less than or equal to F0, it indicates that the current obtained diagnostic recognition rate V is close in size to the historical diagnostic recognition rates, reflecting that the diagnostic recognition rates are less volatile on the time axis and are stable as a whole, so it can be determined that the data acquisition itself is not a problem, at this time the determination criterion can be corrected by reducing the preset diagnostic recognition rate V0, thereby indirectly improving the qualified rate of the fault diagnosis process. If F is greater than F0, it indicates that the historical diagnostic recognition rates obtained on the time axis have a large numerical dispersion, and there may be many extreme values away from the average value, so the data as a whole is unstable, and therefore it can be determined that the current data acquisition has a problem, at this time the diagnosis process can be directly determined to be unqualified, and the difference between V2 and V can be used to determine the reason. The specific value of F0 in the comparison process is determined according to V1 and V2.

[0065] Further, the instruction generation module is further configured to determine to reduce the preset diagnostic recognition rate when the diagnostic recognition variance is less than or equal to the critical diagnostic recognition variance, and to generate a corresponding instruction to determine the reduction amplitude of the preset diagnostic recognition rate based on a comparison result of the recognition variance difference and the preset recognition variance difference, the reduction amplitude being in a positive correlation with the recognition variance difference; and the control module is further configured to control the instruction generation module to reduce the preset diagnostic recognition rate based on the instruction; wherein the recognition variance difference is a difference between the critical diagnostic recognition variance and the diagnostic recognition variance.

[0066] Specifically, in the embodiment, the instruction generation module obtains the critical diagnostic recognition variance F0 and the diagnostic recognition variance F, and further calculates the recognition variance difference B, and more specifically, reduces the first preset diagnostic recognition rate V1, but V1 cannot cross V2 and V1 is always greater than V2; the preset recognition variance difference B0 can be divided into a first preset recognition variance difference B1 and a second preset recognition variance difference B2, B1∈[0.2, 0.3] and B2∈[0.5, 0.7], and the comparison of B1 and B2 with B can improve the adjustment accuracy of the V1 reduction amplitude; the comparison process of B with B1 and B2 is as follows:

[0067] If B is less than or equal to B1, the instruction generation module generates a first recognition rate adjustment coefficient, and the control module controls the instruction generation module to reduce the value range of V1 to [97.8%, 94.8%] based on the first recognition rate adjustment coefficient; if B is greater than B1 and less than or equal to B2, the instruction generation module generates a second recognition rate adjustment coefficient, and the control module controls the instruction generation module to reduce the value range of V1 to [97.5%, 94.5%] based on the second recognition rate adjustment coefficient; if B is greater than B2, the instruction generation module generates a third recognition rate adjustment coefficient, and the control module controls the instruction generation module to reduce the value range of V1 to [97.2%, 94.2%] based on the third recognition rate adjustment coefficient. At this time, B1 and B2 are specific values.

[0068] See also Figure 4 As shown, it is a logical decision diagram for determining the reason for the failure of the diagnostic process and the corresponding processing based on the diagnostic recognition difference in this embodiment. The instruction generation module is also used to determine the reason for the failure of the diagnostic process of the current diagnostic model based on the comparison result of the diagnostic recognition difference and the critical diagnostic recognition difference; the instruction generation module is also used to generate an instruction to issue a notification to optimize the diagnostic model based on the reason, or to obtain the total amount of data based on the determination result and compare it with the critical total amount of data to re-determine the processing; wherein, the diagnostic recognition difference is the difference between the preset diagnostic recognition rate and the diagnostic recognition rate, and the total amount of data is the total amount of data of a single data in the current monitoring cycle.

[0069] Specifically, in this embodiment, the preset diagnostic recognition rate V0 is more specifically the second preset diagnostic recognition rate V2, the diagnostic recognition difference N is the difference between V2 and V, and the critical diagnostic recognition difference N0∈[2%,2.5%] is set. N0 is used to determine the size of the obtained N, and then determine the degree of difference between V and V2, and determine the corresponding processing according to the degree of difference; the comparison process based on the diagnostic recognition difference N and N0 is as follows:

[0070] If N is less than or equal to N0, it indicates that the difference between the current V2 and V is relatively small, the total amount of single data can be obtained and recorded as data total amount H, a critical data total amount H0 is set, H and H0 are compared to further determine whether there is a problem in data acquisition or data preprocessing, and then the corresponding processing is determined after the corresponding problem is determined. In the embodiment of the application, H0 can be set as H0∈[13.5GB, 15GB], but the value range of H0 is not limited specifically, and in the optional embodiment, the value range of H0 can also be set as other required value range, for example, H0∈[14GB, 16GB]. If N is greater than N0, it indicates that the difference between the current V2 and V is relatively large, and at this time it is determined that there is a problem in the diagnostic model, which leads to the situation that N is greater than N0. The control module can directly issue a notification to optimize the diagnostic model. N0 is a specific value in the comparison process.

[0071] Further, the instruction generation module is further configured to determine to increase the sampling frequency when the data total amount is less than or equal to the critical data total amount.

[0072] Specifically, in the embodiment, if the data total amount H is less than or equal to the critical data total amount H0, it indicates that the data quantity in the current data preprocessing process is within the range that can be accepted by the current data processing module, and it is determined that there is a problem in the data acquisition process. At this time, the identification difference ratio T is obtained by calculating the ratio of the diagnostic identification difference N and the critical diagnostic identification difference N0, and the increase amplitude of the sampling frequency during data acquisition is determined by the identification difference ratio T. At this time, T is at most 1.

[0073] Further, the instruction generation module is further configured to generate a corresponding instruction to determine the increase amplitude of the sampling frequency based on the comparison result of the identification difference ratio and the preset identification difference ratio, and the increase amplitude and the identification difference ratio are in a negative correlation relationship. The control module is further configured to control the data acquisition module to increase the sampling frequency based on the instruction. The identification difference ratio is the ratio of the diagnostic identification difference and the critical diagnostic identification difference.

[0074] Specifically, in the embodiment, the preset identification difference ratio T0 can be divided into a first preset identification difference ratio T1 and a second preset identification difference ratio T2, T1∈[0.2, 0.3] and T2∈[0.4, 0.5] are set, and T1 and T2 are compared with the identification difference ratio T to refine the increase amplitude of the sampling frequency. The comparison process based on T, T1 and T2 is as follows:

[0075] If T is less than or equal to T1, the instruction generation module generates a first sampling frequency adjustment coefficient, and the control module controls the data acquisition module to increase the original sampling frequency by 35% based on the first sampling frequency adjustment coefficient. If T is greater than T1 and less than or equal to T2, the instruction generation module generates a second sampling frequency adjustment coefficient, and the control module controls the data acquisition module to increase the original sampling frequency by 30% based on the second sampling frequency adjustment coefficient. If T is greater than T1, the instruction generation module generates a third sampling frequency adjustment coefficient, and the control module controls the data acquisition module to increase the original sampling frequency by 20% based on the third sampling frequency adjustment coefficient. The dynamic increase adjustment of the sampling frequency is a prior art and will not be described in detail. In the embodiment of the present application, the increase multiple of the sampling frequency is not specifically limited, and the increase multiple can also be set to other values that meet the requirements. For example, when T is less than or equal to T1, the sampling frequency can also be increased by 40% based on the initial value. It should be noted that the increased sampling frequency should not exceed the sampling frequency increase standard requirement to avoid excessive sampling due to the effective signal being overwhelmed by noise. If the increased sampling frequency does not meet the requirement, the increase multiple is re-determined.

[0076] Further, the instruction generation module is further configured to determine to increase the block quantity of the data processing module in the preprocessing process when the total data quantity is greater than the critical total data quantity.

[0077] Specifically, in the embodiment, if the total data quantity H is greater than the critical total data quantity H0, it indicates that the current data quantity is relatively large, and the block quantity in the current preprocessing process can not meet the requirement of timely processing of data, thereby causing the diagnostic recognition rate V to be less than the second preset diagnostic recognition rate V2. At this time, it is determined that there is a problem in the data preprocessing process. The difference K between H and H0 is obtained by calculating the difference between H and H0, and the increase amplitude of the block quantity is determined by K.

[0078] Further, the instruction generation module is further configured to generate a corresponding instruction to determine the increase amplitude of the block quantity based on a comparison result of the total data quantity difference and a preset total data quantity difference, the increase amplitude and the total data quantity difference are in a positive correlation relationship; the control module is further configured to control the data processing module to increase the block quantity based on the instruction; wherein the total data quantity difference is a difference between the total data quantity and the critical total data quantity.

[0079] Specifically, in the embodiment, a 1MB size block can be adopted, the preset data total amount difference K0 can be divided into a first preset data total amount difference K1 and a second preset data total amount difference K2, K2 and K1 are compared with the data total amount difference K to refine the increase range of the block quantity, K1 is set to be in [0.2GB, 0.3GB], and K2 is set to be in [0.5GB, 0.6GB]; the comparison process of K and K1 and K2 is specifically as follows:

[0080] If K is less than or equal to K1, the instruction generation module generates a first block quantity adjustment coefficient, and the control module controls the data processing module to increase 3 more than the original block quantity based on the first block quantity adjustment coefficient. If K is greater than K1 and less than or equal to K2, the instruction generation module generates a second block quantity adjustment coefficient, and the control module controls the data processing module to increase 6 more than the original block quantity based on the second block quantity adjustment coefficient. If K is greater than K2, the instruction generation module generates a third block quantity adjustment coefficient, and the control module controls the data processing module to increase 9 more than the original block quantity based on the third block quantity adjustment coefficient. The dynamic increase adjustment of the block quantity belongs to the prior art and will not be described in detail. K1 and K2 at this time are specific values, and in the embodiment, the increase quantity is not specifically limited, and in an optional embodiment, the increase quantity can also be set to other values that meet the requirements, for example, when K is greater than K2, the block quantity can also increase 10 more than the original quantity.

[0081] Further, an alarm module and a maintenance notification module are further included; the alarm module is connected with the fault diagnosis module and the instruction generation module respectively, to generate an alarm level report based on the monitoring result in the case that the diagnosis process of the diagnosis model is qualified; the maintenance notification module is connected with the alarm module, to determine to send a notification of a corresponding maintenance scheme based on the alarm level report.

[0082] Please refer to Figure 1As shown, the train intelligent monitoring system based on the Internet of Things sensor in the embodiment further comprises an alarm module and a maintenance notification module. In the case where it is determined that the diagnosis process of the diagnosis model is qualified, the monitoring result can be directly sent to the alarm module, and an alarm level report is generated by the alarm module, wherein the alarm level report comprises: a warning level (annotated in orange): at least two parameters exceed the threshold value or 3% of a single parameter exceeds the threshold value, which affects the normal operation of the train or equipment, and needs to be handled as soon as possible; a danger level (annotated in red): at least two parameters exceed the threshold value or 3% of a single parameter exceeds the threshold value and lasts for at least 3 minutes, which involves train safety or directly endangers life, and needs to be handled immediately. It can be understood that there are various types of abnormal data at one part of the train, for example, for the wheels of the train, the fault types it includes are bearing failure, wheel speed signal anomaly, temperature anomaly, etc., and there are corresponding safety threshold parameters for each fault type; it should be noted that the percentage values of the threshold values corresponding to different parameters are different, so the 3% in the above alarm level report can be adaptively adjusted to other values. After obtaining the alarm level report, the maintenance notification module issues a corresponding maintenance scheme notification, and the maintenance personnel execute the instructions according to the predetermined maintenance scheme, wherein the predetermined maintenance scheme comprises: when in the warning level, continuously collect fault data, wait for the train to be overhauled, and use the fault data as the basis for overhaul and record the overhaul process to improve the efficiency of the overhaul; when in the warning level, continuously collect multi-source data of the corresponding part and re-diagnose the fault, if the subsequent monitoring result does not rise to the danger level, the corresponding part is determined to have a corresponding fault, and the train is directed to the nearest repair point for repair and the repair process is recorded; when in the danger level, it means that the corresponding key part has a failure risk, and the train needs to be parked immediately and the staff needs to be arranged for repair and the repair process needs to be recorded, and after initially ensuring that the train can run, the train is directed to the nearest repair point for comprehensive repair and the repair process is recorded.

[0083] It can be understood that in the embodiment of the application, any one of the preset parameters or the critical parameters is not specifically limited, the preset parameters include a preset diagnosis recognition rate, a preset recognition difference ratio, a preset data total amount difference, and the critical parameters include a critical diagnosis recognition variance, a critical diagnosis recognition difference, and a critical data total amount. The above-mentioned any value is not limited thereto, and a person skilled in the art can adjust the preset parameters and the critical parameters according to actual needs or historical data analysis.

[0084] In order to better illustrate the determination of the diagnosis process of the diagnosis model and the corresponding processing, the following two specific embodiments are described.

[0085] Embodiment one

[0086] In the embodiment, the system of the application takes the collected gear crack signal data as an example to perform fault diagnosis analysis to output monitoring results. When the diagnosis is analyzed through the historical operation data and the historical diagnosis data, a diagnosis recognition rate V=94.9% is obtained. At this time, V is less than a first preset diagnosis recognition rate V1=95% and greater than or equal to a second preset diagnosis recognition rate V2=92%. Therefore, the historical diagnosis recognition rates of the corresponding gears and the current diagnosis recognition rate need to be obtained to perform re-determination. The diagnosis recognition variance F=0.25 is obtained by performing variance calculation on the obtained diagnosis recognition rates. At this time, F belongs to the case that is less than a critical diagnosis recognition variance F0. It can be determined that there is no problem in the acquisition of the data itself. At this time, the determination criterion can be corrected by reducing the first preset diagnosis recognition rate V1, and the new determination criterion is used to perform re-determination. F0 takes a value of 0.3. The recognition variance difference B=0.05 is obtained by calculating F0 and F. At this time, B belongs to the case that is less than B1. The first recognition rate adjustment coefficient L=0.02% can be generated by the instruction generation module. The new V1=94.8% is determined by subtracting L from the original V1. The probability of V being greater than V1 is increased when re-determination is performed, so as to output more qualified monitoring results and improve the diagnosis efficiency.

[0087] Embodiment Two

[0088] In the embodiment, the system of the application takes the collected wheel bearing vibration signal data as an example to perform fault diagnosis analysis to output monitoring results. When the diagnosis is analyzed through the historical operation data and the historical diagnosis data, a diagnosis recognition rate V=90.2% is obtained. At this time, V is less than the second preset diagnosis recognition rate V2=93%. It can be determined that the diagnosis process of the diagnosis model at this time is unqualified. The diagnosis recognition difference N=2.8% is obtained by calculating V and V2. At this time, N belongs to the case that is greater than a preset diagnosis recognition difference N0. It is indicated that the current vibration recognition rate V and the minimum required V2 are greatly different. It is directly determined that there is a problem in the diagnosis model. At this time, the control module sends a notification to optimize the current model. The new intelligent optimization algorithm can be combined with the model.

[0089] Thus, the technical solutions of the application have been described in connection with the preferred embodiments illustrated in the drawings. However, those skilled in the art can easily understand that the protection scope of the application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the related technical features without departing from the principles of the application. The technical solutions after the changes or replacements will fall within the protection scope of the application.

[0090] The above merely illustrates the preferred embodiments of the present application, and is not used to limit the present application; for those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A train intelligent monitoring system based on Internet of Things sensors, characterized in that: include: A data acquisition module, which includes a number of sensors for collecting data from corresponding parts of the train; a data processing module connected to the data acquisition module, configured to receive the data and perform preprocessing to obtain a preprocessed data set; a fault diagnosis module connected to the data processing module, configured to obtain the pre-processed data set and perform fault diagnosis analysis according to a diagnostic model to output a monitoring result; a database comprising a plurality of historical operating data and a plurality of historical diagnostic data, connected to the fault diagnosis module and used to transmit the plurality of historical operating data to the fault diagnosis module to generate an operating result; an analysis and processing module, connected to the fault diagnosis module and the database, respectively, for comparing a number of the historical diagnosis data with the operation results to determine the diagnosis recognition rate of the current diagnosis model; an instruction generation module connected to the analysis and processing module, configured to determine whether the diagnostic process of the diagnostic model is qualified based on the diagnostic recognition rate, and to obtain a number of historical diagnostic recognition rates based on the determination result to re-determine whether the diagnostic process is qualified, or to generate an instruction if the diagnostic process is unqualified, the instruction being to re-determine the preset diagnostic recognition rate in the instruction generation module, the sampling frequency in the data acquisition module, or the number of blocks in the data processing module; A control module is connected to the instruction generation module, the data acquisition module and the data processing module respectively, and is used to determine and adjust the preset diagnostic recognition rate, the sampling frequency or the number of blocks based on the instruction.

2. The train intelligent monitoring system based on Internet of Things sensors according to claim 1 is characterized in that: The instruction generation module is further configured to determine whether the diagnostic process of the diagnostic model is qualified based on a comparison result of the diagnostic recognition rate and the preset diagnostic recognition rate, and, based on the determination result, obtain a plurality of the historical diagnostic recognition rates of the corresponding part and a current diagnostic recognition rate to calculate a diagnostic recognition variance and re-determine whether the diagnostic process is qualified; The instruction generation module is further configured to determine a reason for the failure based on a difference between the preset diagnosis recognition rate and the diagnosis recognition rate when it is determined that the diagnosis process of the diagnosis model is unqualified.

3. The train intelligent monitoring system based on Internet of Things sensors according to claim 2 is characterized in that: The instruction generation module is further configured to determine whether to reduce the preset diagnosis recognition rate based on a comparison result of the diagnosis recognition variance and a critical diagnosis recognition variance.

4. The train intelligent monitoring system based on Internet of Things sensors according to claim 3 is characterized in that: The instruction generation module is further configured to determine, when the diagnostic identification variance is less than or equal to the critical diagnostic identification variance, to reduce the preset diagnostic identification rate, and further configured to generate a corresponding instruction based on a comparison result of the identification variance difference value and the preset identification variance difference value to determine a reduction magnitude of the preset diagnostic identification rate, wherein the reduction magnitude is positively correlated with the identification variance difference value; The control module is further configured to control the instruction generation module to reduce the preset diagnosis recognition rate based on the instruction; The identification variance difference is the difference between the critical diagnosis identification variance and the diagnosis identification variance.

5. The train intelligent monitoring system based on Internet of Things sensors according to claim 2 is characterized in that: The instruction generation module is further configured to determine a reason why the current diagnostic process of the diagnostic model fails based on a comparison result of the diagnostic identification difference value and the critical diagnostic identification difference value; The instruction generation module is further configured to generate an instruction for issuing a notification for optimizing the diagnostic model based on the cause, or to obtain a total amount of data based on a determination result and compare it with a critical total amount of data to redetermine processing; The diagnostic recognition difference is the difference between the preset diagnostic recognition rate and the diagnostic recognition rate, and the total amount of data is the total amount of a single piece of data in the current monitoring period.

6. The train intelligent monitoring system based on Internet of Things sensors according to claim 5 is characterized in that: The instruction generation module is further configured to determine to increase the sampling frequency when the total amount of data is less than or equal to the critical total amount of data.

7. The train intelligent monitoring system based on Internet of Things sensors according to claim 6 is characterized in that: The instruction generation module is further configured to generate a corresponding instruction based on a comparison result of the recognition difference ratio with a preset recognition difference ratio to determine an increase range of the sampling frequency, wherein the increase range is negatively correlated with the recognition difference ratio; The control module is further configured to control the data acquisition module to increase the sampling frequency based on the instruction; The recognition difference ratio is the ratio of the diagnostic recognition difference to the critical diagnostic recognition difference.

8. The train intelligent monitoring system based on Internet of Things sensors according to claim 5 is characterized in that: The instruction generation module is further configured to determine, when the total amount of data is greater than the critical total amount of data, to increase the number of blocks in the preprocessing process of the data processing module.

9. The train intelligent monitoring system based on Internet of Things sensors according to claim 8 is characterized in that: The instruction generation module is further configured to generate a corresponding instruction based on a comparison result of the total data amount difference with a preset total data amount difference to determine an increase in the number of blocks, wherein the increase is positively correlated with the total data amount difference; The control module is further configured to control the data processing module to increase the number of blocks based on the instruction; The total data amount difference is the difference between the total data amount and the critical total data amount.

10. The train intelligent monitoring system based on Internet of Things sensors according to claim 1 is characterized in that: It also includes an alarm module and a maintenance notification module; The alarm module is connected to the fault diagnosis module and the instruction generation module respectively, and is used to generate an alarm level report based on the monitoring result when it is determined that the diagnosis process of the diagnosis model is qualified; The maintenance notification module is connected to the alarm module and is used to determine and issue a notification of a corresponding maintenance plan based on the alarm level report.

Images