阅读说明：本技术 一种电液伺服阀智能故障诊断方法 (Intelligent fault diagnosis method for electro-hydraulic servo valve ) 是由 丁建军 金瑶兰 王洪伦 方群 陆军 于 2021-08-31 设计创作，主要内容包括：本发明涉及一种电液伺服阀智能故障诊断方法,能够快速准确地定位故障点,具体步骤为：1)数据采集：采集电液伺服阀的一些主要参数包括电流、油液温度、压力、阀芯位移；2)建立电液伺服阀正常工作时的数学模型：模型内容主要为伺服阀正常工作时的各种变量的状态值或一个允许的范围；3)在2)步骤的基础上再建立几种伺服阀典型故障的数学模型,这样利于快速甄别出频率较高的伺服阀故障；4)将采集的实时伺服阀参数处理,生成实时模型,并将之与步骤2)建立的模型进行分析对比,若模型与标准模型差异很大,则再将模型与典型故障模型去对比,经过系统的分析对比,得出结论。本发明建立的模型能够有效的提升检测伺服阀故障的速度以及精度。(The invention relates to an intelligent fault diagnosis method for an electro-hydraulic servo valve, which can quickly and accurately locate a fault point and comprises the following specific steps: 1) data acquisition: some main parameters of the electro-hydraulic servo valve are collected, including current, oil temperature, pressure and valve core displacement; 2) establishing a mathematical model of the electro-hydraulic servo valve during normal work: the model content mainly comprises state values or an allowable range of various variables when the servo valve works normally; 3) establishing mathematical models of typical faults of several servo valves on the basis of the step 2), so that faults of the servo valves with high frequency can be screened out quickly; 4) processing the collected real-time servo valve parameters to generate a real-time model, analyzing and comparing the real-time model with the model established in the step 2), comparing the model with a typical fault model if the model is greatly different from the standard model, and obtaining a conclusion through the analysis and comparison of the system. The model established by the invention can effectively improve the speed and the precision of detecting the fault of the servo valve.)

1. An intelligent fault diagnosis method for an electro-hydraulic servo valve can quickly and accurately locate a fault point, and is characterized by comprising the following specific steps:

step 1: collecting key parameters of the electro-hydraulic servo valve;

step 2: establishing a reference mathematical model according to the state of the parameter when the electro-hydraulic servo valve normally works, wherein the reference mathematical model integrates the variation characteristics of each variable factor and the reference of the parameter state when the electro-hydraulic servo valve actually runs for many times, so that the reference mathematical model has ultrahigh flexibility and ultrahigh generalization capability, can quickly discriminate the acquired abnormal state parameter, and further accurately analyzes the fault location; in addition, in order to rapidly process data, a parameter training set is established on the basis of a reference model and is used for rapidly sorting acquired parameters, so that the fault diagnosis efficiency is improved; the reference mathematical model is used as a reference template to be compared with the generated electro-hydraulic servo valve real-time mathematical model for analysis and operation, so that the aim of rapid diagnosis is fulfilled;

and step 3: and 2, establishing several typical fault mathematical models of the servo valve on the basis of the step 2, and classifying the faults into typical faults and atypical faults according to the types of the faults, wherein the typical faults comprise working edge wear and valve core clamping stagnation, the atypical faults comprise too high and too low oil temperature, too high and too low oil pressure,

temperature sensor, pressure sensor failure, valve coil open circuit;

and 4, step 4: parameter processing, namely establishing a real-time mathematical model, performing parameter training on key data acquired from an electro-hydraulic servo valve, and rapidly establishing the real-time mathematical model by training the acquired data so as to compare the real-time mathematical model with the established reference mathematical model and the established typical fault mathematical model;

and 5: and (4) analyzing and comparing the real-time mathematical model generated in the step (4) with the established reference mathematical model and the established typical fault mathematical model, and finally obtaining a conclusion.

2. The intelligent fault diagnosis method for the electro-hydraulic servo valve according to claim 1, characterized in that:

the key parameters of the electro-hydraulic servo valve collected in the step 1 comprise: the method comprises the steps of feeding back current, oil temperature, pressure, valve core displacement and the like, wherein the parameters are parameters participating in establishing a working mathematical model of the servo valve, and the states of the parameters can display the working state of the servo valve to a certain extent.

3. The intelligent fault diagnosis method for the electro-hydraulic servo valve according to claim 1, characterized in that:

the typical fault mathematical model established in the step3 diagnoses the fault of the servo valve, adopts a matching matrix K which is obtained by deduction, and uses the matching matrix K to perform matching analysis operation, and the specific process is as follows:

step3.1: the matching matrix K is solved,

[NWM]_m×n＝[K]_m×n[TFM]_n×n

[NWM]_m×nis a state matrix, [ TFM ] of the servo valve in normal operation]_n×nIs a matrix of states at a typical fault,

step3.2: the contents of the matching array K are fault matched,

f₁,f₂,...,f_m,f_na matrix matched for each fault condition.

4. The intelligent fault diagnosis method for the electro-hydraulic servo valve according to claim 1, characterized in that:

step4, processing the acquired servo valve parameter data, wherein the process is as follows:

step4.1: noise reduction processing is carried out on the servo valve sampling data, the collected servo valve parameter data form a matrix A with m rows and n columns, singular value decomposition is carried out, and the decomposition is as follows:

wherein U represents the similar direction among the dimensional data, V shows the similar degree among each data, Σ is a diagonal matrix, the value on the diagonal is a singular value, the number of non-zero singular values is the rank of the matrix, T is a transposed symbol, m, n are integers greater than 1;

when the selected dimension data are correlated, the singular value has a zero value; if not, the singular values are all non-zero values; if the selected dimension data are not related and the difference between singular values is large, noise is considered to exist, the singular values smaller than the data threshold are reset to zero by setting the data threshold, and the noise data can be eliminated by synthesizing the matrix again;

step4.2: processing several main parameters after noise reduction,

Δ is the normalized spool displacement, p is the normalized pressure, T is the normalized temperature, I is the normalized current, Δ_sTo sample the displacement, p_sFor sampling pressure, T_sTo sample the temperature, I_sTo sample the current, Δ_maxFor maximum spool displacement, Δ_cFor a given spool displacement, p_maxIs the maximum pressure, p_gFor a given pressure, T_maxUpper limit of temperature, T_eIs ambient temperature, I_cFor a given temperature;

step4.3: the real-time mathematical model is solved,

in order to be a sampling matrix, the sampling matrix,

standardizing a matrix for the model;

[R-tM]_n×nis a real-time model matrix.

5. The intelligent fault diagnosis method for the electro-hydraulic servo valve according to claim 1, characterized in that:

step5 is to compare the real-time mathematical model with the typical fault mathematical model, and the process is as follows:

step5.1: and (3) solving a similarity matrix k of the real-time mathematical model and the typical fault mathematical model:

[NWM]_m×n＝[k]_m×n[R-tM]_n×n

step5.2: the matrix K is compared to the matching matrix K,

contrast matching arrayWhere the similarity to matrix k is very high, the fault point can be located.

Technical Field

The invention relates to a fault diagnosis technology of an electro-hydraulic servo valve, in particular to an intelligent fault diagnosis method of the electro-hydraulic servo valve.

Background

The electro-hydraulic servo valve is the central nerve of a hydraulic system executing mechanism, if the electro-hydraulic servo valve can learn, the change difference or the running state difference in the system can be sensed, self-regulation, namely intellectualization, is realized, and the intelligent electro-hydraulic servo valve can be compared with the brain of a person.

At present, an airborne hydraulic system judges the fault of a servo valve only by detecting the position of a valve core through the servo valve with a displacement sensor, and judges the fault of the valve by detecting the change of a pressure sensor or a flow detection device in a pipeline connected with an actuating mechanism (load equipment) without the servo valve with the displacement sensor. Often, the functions of the hydraulic system and the airborne control part are realized by different personnel, and the professional understanding and the provided demand difference cause that the high fusion judgment is difficult to achieve. Particularly, for an airplane airborne electrohydraulic servo valve, in the working process, the valve is in extreme environments of high temperature, high pressure, strong vibration, high dynamics and the like, so that the acquired signals can be greatly interfered, effective information is easily submerged in noise, and the signal acquisition and analysis processing of the electrohydraulic servo valve are extremely difficult.

With the increasing complexity of the aircraft hydraulic system, in order to ensure the safety and reliability of the aircraft hydraulic system, the number of monitoring sensors required by key components such as an electro-hydraulic servo valve and the like is correspondingly increased, so that the weight of the aircraft is increased, and a series of problems such as performance reduction and oil consumption increase of the aircraft are generated.

Disclosure of Invention

The invention aims to provide an intelligent fault diagnosis method for an electro-hydraulic servo valve, which is used for solving the problem of judging the fault of the valve by detecting the change of a pressure sensor or flow detection equipment in a pipeline connected with an actuating mechanism (load equipment), and effectively improving the precision and speed of detecting the fault.

In order to achieve the purpose, the technical scheme of the invention is as follows: an intelligent fault diagnosis method for an electro-hydraulic servo valve can quickly and accurately locate a fault point, and comprises the following specific steps:

step 1: collecting key parameters of the electro-hydraulic servo valve;

step 2: establishing a reference mathematical model according to the state of the parameter when the electro-hydraulic servo valve normally works, wherein the reference mathematical model integrates the variation characteristics of each variable factor and the reference of the parameter state when the electro-hydraulic servo valve actually runs for many times, so that the reference mathematical model has ultrahigh flexibility and ultrahigh generalization capability, can quickly discriminate the acquired abnormal state parameter, and further accurately analyzes the fault location; in addition, in order to rapidly process data, a parameter training set is established on the basis of a reference model and is used for rapidly sorting acquired parameters, so that the fault diagnosis efficiency is improved; the reference mathematical model is used as a reference template to be compared with the generated electro-hydraulic servo valve real-time mathematical model for analysis and operation, so that the aim of rapid diagnosis is fulfilled;

and step 3: establishing a plurality of servo valve typical fault mathematical models on the basis of the step 2, and classifying the faults into typical faults and non-typical faults according to the types of the faults, wherein the typical faults comprise working edge abrasion and valve core clamping stagnation, the atypical faults comprise overhigh and overlow oil temperature, overhigh and overlow oil pressure, faults of a temperature sensor and a pressure sensor and open circuit of a valve coil;

and 4, step 4: parameter processing, namely establishing a real-time mathematical model, performing parameter training on key data acquired from an electro-hydraulic servo valve, and rapidly establishing the real-time mathematical model by training the acquired data so as to compare the real-time mathematical model with the established reference mathematical model and the established typical fault mathematical model;

and 5: and (4) analyzing and comparing the real-time mathematical model generated in the step (4) with the established reference mathematical model and the established typical fault mathematical model, and finally obtaining a conclusion.

Further, the key parameters of the electro-hydraulic servo valve collected in the step 1 include: the method comprises the steps of feeding back current, oil temperature, pressure, valve core displacement and the like, wherein the parameters are parameters participating in establishing a working mathematical model of the servo valve, and the states of the parameters can display the working state of the servo valve to a certain extent.

Further, the typical fault mathematical model established in the step3 diagnoses the fault of the servo valve, a matching matrix K is adopted, and the matching matrix K is used for performing matching analysis operation, and the specific process is as follows:

step3.1: the matching matrix K is solved,

[NWM]_m×n＝[K]_m×n[TFM]_n×n

[NWM]_m×nis a state matrix, [ TFM ] of the servo valve in normal operation]_n×nIs a matrix of states at a typical fault,

step3.2: the contents of the matching array K are fault matched,

f₁,f₂,...,f_m,f_na matrix matched for each fault condition.

Further, the step4 processes the acquired servo valve parameter data, and the process is as follows:

step4.1: noise reduction processing is carried out on the servo valve sampling data, the collected servo valve parameter data form a matrix A with m rows and n columns, singular value decomposition is carried out, and the decomposition is as follows:

wherein U represents the similar direction among the dimensional data, V shows the similar degree among each data, Σ is a diagonal matrix, the value on the diagonal is a singular value, the number of non-zero singular values is the rank of the matrix, T is a transposed symbol, m, n are integers greater than 1;

when the selected dimension data are correlated, the singular value has a zero value; if not, the singular values are all non-zero values; if the selected dimension data are not related and the difference between singular values is large, noise is considered to exist, the singular values smaller than the data threshold are reset to zero by setting the data threshold, and the noise data can be eliminated by synthesizing the matrix again;

step4.2: processing several main parameters after noise reduction,

Δ is the normalized spool displacement, p is the normalized pressure, T is the normalized temperature, I is the normalized current, Δ_sTo sample the displacement, p_sFor sampling pressure, T_sTo sample the temperature, I_sTo sample the current, Δ_maxFor maximum spool displacement, Δ_cFor a given spool displacement, p_maxIs the maximum pressure, p_gFor a given pressure, T_maxUpper limit of temperature, T_eIs ambient temperature, I_cFor a given temperature;

step4.3: the real-time mathematical model is solved,

in order to be a sampling matrix, the sampling matrix,

standardizing a matrix for the model;

[R-tM]_n×nis a real-time model matrix.

Further, the step5 compares the real-time mathematical model with the typical fault mathematical model, and the process is as follows:

step5.1: and (3) solving a similarity matrix k of the real-time mathematical model and the typical fault mathematical model:

[NWM]_m×n＝[k]_m×n[R-tM]_n×n

step5.2: the matrix K is compared to the matching matrix K,

contrast matching arrayWhere the similarity to matrix k is very high, the fault point can be located.

The invention has the advantages and beneficial effects that:

1. the invention provides an idea for establishing a real-time model of a servo valve, which comprises the following steps: the sampled data is standardized to build a real-time model of the servo valve. Compared with other models, the model has the advantages of less training parameters and high training speed, and the efficiency and the precision of fault diagnosis of the electro-hydraulic servo valve are effectively improved.

2. The invention can be used in the aviation field, can be also used in other industrial fields with higher automation degree and convenient information acquisition, and has larger practical application value.

Drawings

FIG. 1 is a block diagram of the structure of the intelligent fault diagnosis method for the electro-hydraulic servo valve of the present invention;

FIG. 2 is a flow diagram of a fault handling model;

FIG. 3 is a model diagnostic flow chart.

Detailed Description

The invention will be further explained with reference to the drawings.

As shown in fig. 1 to 3, the intelligent fault diagnosis method for an electro-hydraulic servo valve of the present invention specifically includes the following steps:

in the first step, key parameters of the electro-hydraulic servo valve, including temperature, pressure, current and valve core displacement, are acquired.

And secondly, carrying out standardization processing on the acquired key parameters, and generating a real-time mathematical model of the electro-hydraulic servo valve by using the processed parameters.

And thirdly, matching the real-time mathematical model with the established typical fault mathematical model.

And fourthly, outputting the fault type according to the operation result of the third step.

Firstly, introducing a typical fault mathematical model established in the third step, wherein the specific steps are as follows:

the typical fault mathematical model established in the step3 needs to diagnose the fault of the electro-hydraulic servo valve, a matching matrix K is deduced without directly using the fault model for matching operation, and the matrix K is used for matching analysis operation, and the specific process is as follows:

step3.1: the matching matrix K is solved,

[NWM]_m×n＝[K]_m×n[TFM]_n×n

[NWM]_m×nis a state matrix, [ TFM ] of the servo valve in normal operation]_n×nIs a matrix of states at a typical fault,

step3.2: the contents of the matching array K are fault matched,

f₁,f₂,...,f_m,f_nfor each matrix of fault state matches,

further, the fourth step of data processing includes the following specific steps:

step4.1: carrying out noise reduction processing on the electro-hydraulic servo valve sampling data, forming the acquired electro-hydraulic servo valve parameter data into a matrix A with m rows and n columns, and carrying out singular value decomposition as follows:

wherein U represents the similar direction among the dimensional data, V shows the similar degree among each data, Σ is a diagonal matrix, the value on the diagonal is a singular value, the number of non-zero singular values is the rank of the matrix, T is a transposed symbol, m, n are integers greater than 1;

when the selected dimension data are correlated, the singular value has a zero value; if not, the singular values are all non-zero values; and if the selected dimension data are not related and the difference between the singular values is large, the noise is considered to exist, the singular values smaller than the data threshold are reset to zero by setting the data threshold, and the noise data can be eliminated by synthesizing the matrix again.

Step4.2: the data after the noise reduction is processed,

Δ is the normalized spool displacement, p is the normalized pressure, T is the normalized temperature, I is the normalized current, Δ_sTo sample the displacement, p_sFor sampling pressure, T_sTo sample the temperature, I_sTo sample the current, Δ_maxFor maximum spool displacement, Δ_cFor a given spool displacement, p_maxIs the most importantHigh pressure, p_gFor a given pressure, T_maxUpper limit of temperature, T_eIs ambient temperature, I_cCorresponding to a given temperature.

Step4.3: the real-time mathematical model is solved,

in order to be a sampling matrix, the sampling matrix,

standardizing a matrix for the model;

[R-tM]_n×nas a matrix of real-time models

And step five, comparing the real-time mathematical model with a typical fault mathematical model, wherein the process is as follows:

step5.1: and (3) solving a similarity matrix k of the real-time model and the typical fault mathematical model:

[NWM]_m×n＝[k]_m×n[R-tM]_n×n

step5.2: the matrix K is compared to the matching matrix K,

contrast matching arrayWhere the similarity to matrix k is very high, the fault point can be located.