Lubricating equipment component fault prediction test system

Abstract

The invention discloses a lubricating equipment component fault prediction test system, which relates to the technical field of lubricating equipment state monitoring and comprises a multi-source sensing module, a particle feature resolving module, an entropy analysis module, a collaborative diagnosis module and a decision execution module. According to the lubricating equipment part fault prediction test system, by means of the particle feature resolving module, materials of metal and non-metal abrasion particles can be accurately recognized, the flow speed and the concentration of the particles can be calculated, a detailed material proportion, size distribution and a real-time concentration thermodynamic diagram can be generated, and the entropy value analysis module is used for analyzing the real-time temperature and pressure data according to the real-time temperature and pressure data. An infinitesimal body entropy change value is accurately calculated, the phase change condition of a lubricant is accurately judged in combination with a preset threshold value, a collaborative diagnosis module carries out deep fusion and analysis on multi-source data, time asynchronism is eliminated through dynamic time warping and other algorithms, an association rule base is established, confidence is calculated, and therefore the fault type of a lubricating equipment component is comprehensively predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of lubrication equipment status monitoring, in particular to a lubrication equipment component failure prediction test system. Background Art

[0002] The lubrication system is an important component of key transmission components such as bearings and gearboxes. Its status directly determines the life and operational safety of the equipment. As the operating speed and load of the equipment increase, higher requirements are placed on the performance of the lubrication system. Failure of lubrication equipment components will cause equipment downtime and affect production efficiency. Therefore, predicting failures of lubrication equipment components, detecting potential failures in advance and taking corresponding measures have become key issues that need to be urgently addressed in the industrial field.

[0003] Existing fault detection systems for lubrication equipment components rely on a single type of sensor data, such as using only temperature sensors or vibration sensors to determine the equipment status. This approach cannot fully capture the complex information during the equipment operation process and is prone to misjudgment and omissions. For the detection of wear particles, traditional methods have difficulty accurately identifying their material and size, and cannot deeply analyze the cause and extent of wear. When dealing with the phase change problem of lubricants, existing technologies lack effective entropy analysis methods and cannot timely and accurately determine the phase change state of the lubricant and its relationship with equipment failures. In addition, existing systems have deficiencies in data fusion and collaborative diagnosis, and cannot fully utilize multi-source data for comprehensive analysis, resulting in limited accuracy and timeliness of fault diagnosis.

[0004] In summary, the existing fault detection technology for lubrication equipment components has limitations and cannot meet the requirements of modern industry for equipment reliability and maintenance efficiency. Therefore, there is an urgent need for a fault prediction test system that can comprehensively utilize multi-source data, accurately identify wear particle characteristics, deeply analyze lubricant phase changes, and achieve efficient collaborative diagnosis. Summary of the Invention

[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide a lubrication equipment component fault prediction test system, which can realize the online sorting of metal / non-metallic particles by integrating multi-frequency eddy current particle sorting and multiphase flow entropy analysis, dynamically capture the lubricant phase change critical point by combining entropy change data, and construct a particle-entropy change association rule library, so as to comprehensively and accurately predict the fault type of lubrication equipment components.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a lubrication equipment component fault prediction test system, the system comprising: a multi-source perception module, a particle feature solution module, an entropy analysis module, a collaborative diagnosis module, and a decision execution module;

[0007] The multi-source sensing module integrates an annular multi-frequency eddy current sensor group in the lubrication pipeline to sense the eddy current changes caused by wear particles in the pipeline passing through the alternating magnetic field. It is also equipped with a temperature sensor and a pressure sensor to sense the temperature and pressure data of the lubricant in real time.

[0008] The particle feature calculation module receives signals from the multi-frequency eddy current sensor group in the multi-source perception module, identifies the materials of metal and non-metal wear particles by calculating the eddy current phase difference at different frequencies, and calculates the particle flow rate and concentration based on the particle motion trajectory information to generate a real-time concentration heat map of the material distribution;

[0009] The entropy value analysis module calculates the entropy change value of the microelement based on the real-time temperature and pressure data of the multi-source sensing module, and makes a judgment based on the preset entropy thresholds of the lubricant solid-liquid phase change and liquid-gas phase change;

[0010] The collaborative diagnosis module uses dynamic time warping to align particle data with entropy change time series, eliminates time asynchrony, establishes an association rule base and calculates confidence, and outputs the fault type and confidence score;

[0011] The decision execution module issues an early warning signal according to the diagnosis result of the collaborative diagnosis module.

[0012] Furthermore, the particle feature solving module selects two eddy current sensors with different frequencies f1 and f2 in the multi-frequency eddy current sensor group, and the initial output signal phases thereof are φ 01 、φ 02 When wear particles pass through the alternating magnetic field, the phase of the sensor signal changes to φ1 and φ2. The phase difference between the two frequencies is calculated as Δφ1 = φ1-φ 01 ,Δφ2=φ2-φ 02 , which is used to reflect the interference of particles on eddy current signals of different frequencies, thereby establishing the material judgment function M(Δφ1,Δφ2);

[0013] For metal particles, the material judgment function is M js (Δφ1, Δφ2)=a1Δφ1+b1Δφ2+c1;

[0014] For non-metallic particles, the function is M fjs (Δφ1, Δφ2)=a2Δφ1+b2Δφ2+c2;

[0015] Among them, a1, b1, c1 and a2, b2, c2 are judgment coefficients, which are used to quantify the influence of different frequency phase differences on the identification of metal and non-metal particles. The calculated M(Δφ1, Δφ2) value is compared with M js (Δφ1, Δφ2) and M fjs(Δφ1, Δφ2) for comparison, when M (Δφ1, Δφ2)≈M js If (Δφ1, Δφ2) is greater than Δφ1, it is judged as a metal particle, otherwise it is judged as a non-metal particle.

[0016] Furthermore, the particle feature solution module calculates the particle flow rate as follows: at two consecutive moments t1 and t2, with a time interval Δt = t2-t1, the position coordinates of the particle in three-dimensional space are obtained by the position monitoring sensor as (x1, y1, z1) and (x2, y2, z2), and the displacement of the particle in the time Δt is calculated. The particle flow rate Among them, (x1, y1, z1) and (x2, y2, z2) are the spatial position coordinates of the particles at different times. The flow velocity is obtained by the ratio of displacement to time, which intuitively reflects the speed of particle movement in the pipeline.

[0017] Furthermore, the particle feature solution module calculates the particle concentration as follows: the lubrication pipeline is divided into n small sections along the length direction, and any lubrication pipeline area with a length of L is selected from the lubrication pipeline area. The cross-sectional area of the area is S, and the number of particles in the area detected by the multi-frequency eddy current sensor group is N. The volume of the area is V = S × L, then the particle concentration is That is, the number of particles per unit volume, which is used to measure the density of particles in the pipeline. The material, position, flow rate, and concentration of the particles are used as the particle data sequence D, and a real-time concentration heat map of the material distribution is generated.

[0018] Furthermore, the entropy value analysis module calculates the entropy change value of the microelement based on the real-time temperature and pressure data of the multi-source sensing module, that is, Among them, m is the number of temperature sensors, ΔS is the entropy change value of the microelement, q i is the heat flux density between the i-th adjacent temperature measurement points and k is thermal conductivity, T i+1 and T i are the temperature values of two adjacent temperature measuring points, L is the length between the adjacent temperature measuring points, A is the heat transfer area, P is the pressure, ΔV gas is the change in gas volume.

[0019] Furthermore, the process of threshold matching performed by the entropy analysis module is as follows:

[0020] When the real-time entropy change ΔS meets the conditions, the phase change flag F is triggered. phase ∈{0,1,2}, that is, when |ΔS-S SL |<0.05S SL When the phase change flag F is set phase =1 indicates solid-liquid phase transition;

[0021] When |ΔS-S LG |<0.05S LG When the phase change flag F is set phase =2 represents liquid-gas phase transition, where S SL is the entropy of the solid-liquid phase transition of the lubricant, S LG is the entropy of the lubricant liquid-gas phase transition, F phase =0 means no phase change, which is used to judge the phase change state of the lubricant and output the results to the collaborative diagnosis module, including: entropy change value ΔS, phase change flag F phase and time t, that is, {ΔS,F phase ,t}.

[0022] Furthermore, the collaborative diagnosis module uses dynamic time warping to align the particle data with the entropy change time series and finds the best match between the two sequences by bending the time axis, i.e. Among them, DT(D,ΔS) represents the dynamic time warping distance between the particle data sequence D and the entropy change data sequence ΔS. The smaller the value, the higher the similarity between the two sequences. π represents the set of all time warping paths. (i, j) is the point on the path π, corresponding to the element in the i-th region of the particle data sequence D and the j-th element in the entropy change data sequence ΔS. D(i) and ΔS(j) are the values of the particle data sequence and the entropy change data sequence at the corresponding positions, respectively. μ D and μ S are the means of the particle data sequence and the entropy change data sequence, respectively, which are used to normalize the data and eliminate the influence of data magnitude differences. D and σ S They are the standard deviations of the particle data sequence and the entropy change data sequence, respectively. They are also used for normalization to make data with different characteristics comparable. The optimal time warping path is found through dynamic time warping to align the time and space of the particle data and the entropy change data.

[0023] Furthermore, the collaborative diagnosis module establishes an association rule base, determines the type of equipment fault by setting different conditions, and calculates the confidence level of each rule. The fault types include bearing overheating and wear and lubricant decomposition and cavitation. The rules of the association rule base are:

[0024] Bearing overheating wear; when the metal particle concentration C js More than 15% and lasts for 5 minutes, and the phase change mark F phase =1, trigger the rule and calculate the confidence α1 = 0.7·sigmoid(C js / 15%)+0.3·sigmoid(ΔS / S SL), when α1 is closer to 1, it indicates that the reliability of judging the bearing overheating and wear fault of the equipment based on the current data and rules is higher; on the contrary, it indicates that the accuracy of the judgment is questionable and the equipment cannot be determined to have the fault. Among them, α1 is the confidence level of bearing overheating and wear, It is an activation function that maps the input value to between 0 and 1 and quantifies the impact of different factors on the confidence level. x represents the input value, C js is the metal particle concentration, ΔS is the entropy change value, S SL is the entropy of the solid-liquid phase transition of the lubricant;

[0025] Lubricant decomposition cavitation: When the non-metallic particle concentration C fjs More than 300%, and the phase change mark F phase =2, the rule is triggered and the confidence α2 = 0.5·tanh(C fjs / 300%)+0.5·exp(-ΔS / S LG ), where α2 is the lubricant decomposition cavitation confidence, tanh(x) is the hyperbolic tangent function, which is used to map the input value to between -1 and 1 to quantify the influence of non-metallic particle concentration, exp(x) is the exponential function, which is used to describe the influence of entropy change on confidence, C fjs is the concentration of non-metallic particles, S LG is the entropy of the lubricant liquid-gas phase transition;

[0026] When two rules are triggered at the same time, the maximum confidence is taken as the final confidence α final , and output the corresponding fault type F fault ∈{1,2}, that is, α final =max(α1,α2), where α final is the final confidence level, when α final When α1 is maximum, F fault =1, indicating bearing overheating and wear failure, when α final When α2 is maximum, F fault =2, indicating lubricant decomposition cavitation failure.

[0027] Compared with the existing technology, this lubrication equipment component failure prediction test system has the following beneficial effects:

[0028] 1. The lubrication equipment component fault prediction test system of the present invention utilizes a particle feature solution module, which can not only accurately identify the material of metal and non-metallic wear particles, but also calculate the particle flow rate and concentration, and generate detailed material proportions, size distribution and real-time concentration heat maps. The entropy analysis module accurately calculates the entropy change value of the microelement based on real-time temperature and pressure data, and accurately judges the phase change of the lubricant in combination with the preset threshold. The collaborative diagnosis module deeply integrates and analyzes multi-source data, uses algorithms such as dynamic time warping to eliminate time asynchrony, establishes an association rule base and calculates confidence, thereby being able to comprehensively and accurately predict the fault type of lubrication equipment components, reducing the possibility of misjudgment and missed judgment.

[0029] 2. The lubrication equipment component fault prediction test system of the present invention uses a multi-source sensing module to comprehensively collect multi-type data such as wear particles, lubricant temperature and pressure in the lubrication pipeline, and conducts a comprehensive assessment of the equipment status from multiple dimensions. Compared with the fault diagnosis method based on a single data source, this system can capture more subtle fault signs, effectively avoid diagnostic deviations caused by one-sided data, greatly improve the accuracy of fault diagnosis, and provide a solid guarantee for the stable operation of the equipment.

[0030] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0032] Figure 1 This is an operational flow chart of a lubrication equipment component failure prediction test system;

[0033] Figure 2 This is a block diagram of a lubrication equipment component failure prediction test system. DETAILED DESCRIPTION

[0034] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0035] Example 1

[0036] This embodiment describes in detail the specific implementation process of a lubrication equipment component failure prediction test system. Figure 2 As shown in the figure, the system integrates a multi-source perception module, a particle feature solution module, an entropy analysis module, a collaborative diagnosis module, and a decision execution module. Through the collaborative work of each module, the collected data is processed and analyzed to achieve accurate monitoring and early warning of lubrication equipment, providing strong guarantees for the safe operation of lubrication equipment.

[0037] The core of the lubrication system is the lubrication pipeline. In order to accurately obtain the status information of wear particles and lubricant in the lubrication pipeline, the multi-source sensing module integrates an annular multi-frequency eddy current sensor group in the lubrication pipeline. The sensor group consists of multiple eddy current sensors of different frequencies. These sensors are tightly surrounded by the lubrication pipeline to ensure that the alternating magnetic field they generate can effectively cover the fluid area inside the pipeline. When the wear particles in the pipeline flow with the lubricant and pass through the alternating magnetic field, eddy current changes will be triggered. This change contains information about the particles, including material and size. At the same time, in order to comprehensively monitor the status of the lubricant, high-precision temperature sensors and pressure sensors are also equipped. The temperature sensor can sense the temperature of the lubricant in real time, and the pressure sensor synchronously obtains pressure data. These data, together with the data collected by the eddy current sensor, constitute the basic information for monitoring the status of the lubrication system. During actual operation, the multi-frequency eddy current sensor group works continuously and continuously emits an alternating magnetic field into the lubrication pipeline. Once a wear particle passes through the magnetic field area, the phase of the sensor's output signal will change. Two eddy current sensors with different frequencies f1 and f2 are selected, and their initial output signal phases are φ respectively. 01 、φ 02 When the wear particles pass through, the signal phase changes to φ1 and φ2. By calculating the phase difference between the two frequencies Δφ1=φ1-φ 01 , Δφ2=φ2-φ 02 , which can reflect the interference of particles on eddy current signals of different frequencies. These two phase differences will serve as an important basis for subsequent judgment of particle material. The temperature sensor and pressure sensor are also working synchronously. They collect data with the same accuracy and frequency, and transmit the data to subsequent modules in real time, providing the entire system with key information about the real-time status of the lubricant.

[0038] The particle feature calculation module receives the signal from the multi-frequency eddy current sensor group in the multi-source perception module, identifies the wear particle material, calculates the particle flow rate and concentration, and generates a real-time concentration heat map. According to the acquired phase differences Δφ1 and Δφ2, a material judgment function M(Δφ1, Δφ2) is established. For metal particles, the material judgment function is M js (Δφ1, Δφ2)=a1Δφ1+b1Δφ2+c1; for non-metallic particles, the function is M fjs(Δφ1, Δφ2)=a2Δφ1+b2Δφ2+c2, where a1, b1, c1 and a2, b2, c2 are judgment coefficients. They are determined by measuring wear particles of known materials and analyzing the influence of different frequency phase differences on particle identification. In practical applications, the calculated M(Δφ1, Δφ2) value is compared with M js (Δφ1, Δφ2) and M f js(Δφ1,Δφ2) is compared, when M(Δφ1,Δφ2)≈M js (Δφ1, Δφ2), the particle is determined to be a metal particle; otherwise, it is determined to be a non-metallic particle. In order to calculate the particle flow rate, at two consecutive moments t1 and t2, with a time interval Δt = t2-t1, the position monitoring sensor is used to obtain the position coordinates of the particle in three-dimensional space, which are (x1, y1, z1) and (x2, y2, z2) respectively. According to the distance formula between two points in space, the displacement of the particle in the time Δt is calculated. The particle flow rate is obtained by the ratio of displacement to time This flow rate value directly reflects the speed of particle movement in the pipeline, and provides an important basis for analyzing the movement state of particles. When calculating the particle concentration, the lubrication pipeline is divided into n small sections along the length direction. From which any lubrication pipeline area with a length of L is selected, the cross-sectional area of the area is S, and the number of particles in the area detected by the multi-frequency eddy current sensor group is N. Then the volume of the area V = S × L, and the particle concentration It represents the number of particles per unit volume and is used to measure the density of particles in the pipeline. The particle feature solution module integrates information such as particle material, position, flow rate, and concentration into a particle data sequence D, and uses this data to generate a real-time concentration heat map of material distribution. During the heat map generation process, the concentration distribution of particles of different materials in the lubrication pipeline is intuitively displayed through the calculation and analysis of particle concentrations in different areas.

[0039] The entropy analysis module calculates the entropy change value of the microelement based on the real-time temperature and pressure data provided by the multi-source perception module, and judges the phase change state of the lubricant by comparing it with the preset threshold, and calculates the heat flux density between the i-th adjacent temperature measurement points. Where k is the thermal conductivity, T i +1 and T i are the temperature values of two adjacent temperature measurement points, L is the length between adjacent temperature measurement points, thermal conductivity k reflects the ability of the lubricant to conduct heat, and the entropy change value of the microelement is calculated. Here, A is the heat transfer area, P is the pressure, which is collected in real time by the pressure sensor, ΔV gas is the volume change of gas, T avgis the average temperature, which is used to normalize the heat to accurately reflect the entropy change of the microelement. After calculating the entropy change value ΔS of the microelement, it needs to be matched with the preset threshold to determine the phase change state of the lubricant. When the real-time entropy change ΔS meets specific conditions, the phase change flag F is triggered. phase ∈{0,1,2}, specifically, when |ΔS-S SL |<0.05S SL When the phase change flag F is set phase =1, indicating solid-liquid phase transition; when |ΔS-S LG |<0.05S LG When the phase change flag F is set phase =2, indicating liquid-gas phase transition; and when F phase = 0, indicating no phase change, where S SL is the entropy of the solid-liquid phase transition of the lubricant, S LG is the entropy value of the lubricant liquid-gas phase change. The entropy value analysis module converts the entropy change value ΔS, the phase change mark F phase The time t is output to the collaborative diagnosis module to provide important data support for subsequent fault diagnosis.

[0040] The collaborative diagnosis module is responsible for aligning the particle data and entropy change time series, establishing an association rule base and calculating the confidence level to determine the fault type and confidence score. Due to the differences in the acquisition time and frequency of particle data and entropy change data, in order to accurately analyze the association between the two, the dynamic time warping algorithm is used to align the particle data sequence D and the entropy change data sequence ΔS. The best match between the two sequences is found by bending the time axis, that is, Among them, DT(D,ΔS) represents the dynamic time warping distance between the particle data sequence D and the entropy change data sequence ΔS. The smaller this distance is, the higher the similarity between the two sequences is. π represents the set of all time warping paths. (i,j) is the point on the path π, corresponding to the element in the i-th region of the particle data sequence D and the j-th element in the entropy change data sequence ΔS. D(i) and ΔS(j) are the values of the particle data sequence and the entropy change data sequence at the corresponding positions, respectively. μ D and μ S are the means of the particle data sequence and the entropy change data sequence, respectively, which are used to normalize the data and eliminate the influence of data magnitude differences. D and σ SThey are the standard deviations of the particle data sequence and the entropy change data sequence, which are also used for normalization to make data with different characteristics comparable. The optimal time warping path is found through the dynamic time warping algorithm to achieve the spatiotemporal alignment of particle data and entropy change data, laying the foundation for subsequent association analysis. An association rule library is established to determine the possible fault types of the equipment by setting different conditions and calculate the confidence of each rule. We focus on the two fault types of bearing overheating wear and lubricant decomposition cavitation. For bearing overheating wear fault, when the metal particle concentration C js More than 15% and lasts for 5 minutes, and the phase change mark F phase =1 (indicates solid-liquid phase transition), the rule is triggered and the confidence α1 = 0.7·sigmoid(C js / 15%)+0.3·sigmoid(ΔS / S SL ), where α1 is the confidence level of bearing overheat wear, It is an activation function that maps the input value to between 0 and 1 and is used to quantify the impact of different factors on the confidence level. x represents the input value, C js is the metal particle concentration, ΔS is the entropy change value, S SL is the entropy value of the solid-liquid phase change of the lubricant. Through this formula, the influence of the metal particle concentration and the entropy change value on the fault judgment is comprehensively considered to obtain a confidence value reflecting the possibility of the fault. For the lubricant decomposition cavitation fault, when the non-metallic particle concentration C fjs More than 300%, and the phase change mark F phase =2, the rule is triggered and the confidence α2 = 0.5·tanh(C fjs / 300%)+0.5·exp(-ΔS / S LG ). Where α2 is the lubricant decomposition cavitation confidence, tanh(x) is the hyperbolic tangent function, which is used to map the input value to between -1 and 1 to quantify the influence of non-metallic particle concentration, exp(x) is the exponential function, which is used to describe the influence of entropy change value on confidence, C fjs is the concentration of non-metallic particles, S LG is the entropy value of the lubricant liquid-gas phase change. Similarly, this formula combines the two factors of non-metallic particle concentration and entropy change value to calculate the confidence of the fault. When multiple rules are triggered at the same time, in order to determine the final fault type, the maximum confidence is taken as the final confidence α final , that is, α final =max(α1,α2), and output the corresponding fault type F according to the rule corresponding to the maximum confidence fault ∈{1,2}, when α final When α1 is maximum, F fault =1, indicating bearing overheating and wear failure; when α fWhen inal corresponds to the maximum value of α2, F fault =2, indicating lubricant decomposition cavitation failure.

[0041] The decision-making execution module issues corresponding early warning signals based on the diagnosis results of the collaborative diagnosis module. At the same time, it pushes maintenance work orders to relevant maintenance personnel, informing them of abnormal equipment conditions, allowing maintenance personnel to make maintenance preparations in advance and inspect and maintain the equipment in a timely manner to ensure its normal operation.

[0042] In summary, this embodiment demonstrates in detail the complete operation process of a lubrication equipment component fault prediction test system, from the multi-source perception module to collect data, to the particle feature solution module to analyze particle information, the entropy analysis module to determine the lubricant phase change, the collaborative diagnosis module to perform fault diagnosis, and then to the decision execution module to take corresponding measures according to the diagnosis results. The various modules work closely together to form an efficient and accurate fault prediction system. The system can comprehensively and accurately monitor the operating status of lubrication equipment components, timely detect potential faults and issue early warnings, providing strong guarantees for the stable operation and preventive maintenance of the equipment.

[0043] Example 2

[0044] like Figure 1 As shown, based on the first embodiment, the specific process of the fault prediction test system for lubricating equipment components in performing fault prediction is as follows:

[0045] (1) Data collection

[0046] Wear particle data collection: The annular multi-frequency eddy current sensor group operates continuously in the lubrication pipeline, emitting an alternating magnetic field into the pipeline. When wear particles flow through the magnetic field area with the lubricant, they cause eddy current changes, resulting in a phase change in the sensor output signal. These phase differences will be used for subsequent particle material identification.

[0047] Temperature and pressure data acquisition: High-precision temperature sensors and pressure sensors work synchronously to sense the temperature and pressure of the lubricant in real time;

[0048] (2) Particle feature calculation

[0049] Particle material identification: Based on the calculated phase difference, substitute it into the material judgment function to determine the particle material;

[0050] Particle flow rate calculation: At two consecutive moments, the position monitoring sensor is used to obtain the position coordinates of the particles in three-dimensional space. The particle displacement is calculated based on the distance formula between two points in space, and then the particle flow rate is calculated to understand the movement of particles in the pipeline;

[0051] Particle concentration calculation: Divide the lubrication pipeline into multiple small sections of equal size along its length. Select one of the sections and use a multi-frequency eddy current sensor group to detect the number of particles in the section and calculate the volume of the section to obtain the particle concentration, which measures the density of particles in the pipeline.

[0052] Real-time concentration heat map generation: Integrate information on particle material, location, flow rate, and concentration, and use the particle data sequence to generate a real-time concentration heat map of material distribution, visually displaying the concentration distribution of particles of different materials in the pipeline;

[0053] (3) Entropy analysis

[0054] Heat flux density calculation: Based on real-time temperature data, the heat flux density between adjacent temperature measurement points is calculated to reflect its heat conduction capacity;

[0055] Calculation of entropy change of microelement: Calculate the entropy change of microelement by combining the information of heat flux density, heat transfer area, temperature, pressure and gas volume change;

[0056] Phase change judgment: Compare the calculated entropy change value of the microelement with the preset threshold and set the phase change flag;

[0057] (4) Collaborative diagnosis

[0058] Data alignment: Dynamic time warping is used to align granular data sequences with entropy-changing data sequences, finding the optimal time warping path to achieve spatiotemporal alignment of data and eliminate the impact of time asynchrony.

[0059] Association rule judgment: Establish an association rule library and set conditions for different fault types;

[0060] Confidence calculation: Calculate confidence based on different rules, taking into account the impact of particle concentration and entropy change on fault judgment;

[0061] Final fault determination: When multiple rules are triggered simultaneously, the maximum confidence level is taken as the final confidence level, and the fault type is output according to the rule corresponding to the maximum confidence level;

[0062] (5) Decision execution

[0063] Early warning signal: According to the output fault type, corresponding early warning signal is issued and measures are taken.

[0064] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A lubrication equipment component failure prediction test system, characterized in that: The system consists of: multi-source perception module, particle feature solution module, entropy analysis module, collaborative diagnosis module, and decision execution module; The multi-source sensing module integrates an annular multi-frequency eddy current sensor group in the lubrication pipeline to sense the eddy current changes caused by wear particles in the pipeline passing through the alternating magnetic field. It is also equipped with a temperature sensor and a pressure sensor to sense the temperature and pressure data of the lubricant in real time. The particle feature calculation module receives signals from the multi-frequency eddy current sensor group in the multi-source perception module, identifies the materials of metal and non-metal wear particles by calculating the eddy current phase difference at different frequencies, and calculates the particle flow rate and concentration based on the particle motion trajectory information to generate a real-time concentration heat map of the material distribution; The entropy value analysis module calculates the entropy change value of the microelement based on the real-time temperature and pressure data of the multi-source sensing module, and makes a judgment based on the preset entropy thresholds of the lubricant solid-liquid phase change and liquid-gas phase change; The collaborative diagnosis module uses dynamic time warping to align particle data with entropy change time series, eliminates time asynchrony, establishes an association rule base and calculates confidence, and outputs the fault type and confidence score; The decision execution module issues an early warning signal according to the diagnosis result of the collaborative diagnosis module.

2. A lubrication equipment component failure prediction test system according to claim 1, characterized in that: The particle feature solving module selects two eddy current sensors with different frequencies f1 and f2 in the multi-frequency eddy current sensor group, and their initial output signal phases are When wear particles pass through the alternating magnetic field, the phase of the sensor signal changes and becomes Calculate the phase difference between two frequencies Used to reflect the interference of particles on eddy current signals of different frequencies, thereby establishing a material judgment function For metal particles, the material judgment function is For non-metallic particles, the function is Among them, a1, b1, c1 and a2, b2, c2 are judgment coefficients, which are used to quantify the influence of different frequency phase differences on the identification of metal and non-metal particles. Value and and For comparison, when M If the particle is positive, it is determined to be a metal particle; otherwise, it is determined to be a non-metal particle.

3. A lubrication equipment component failure prediction test system according to claim 1, characterized in that: The particle feature solution module calculates the particle flow rate as follows: at two consecutive moments t1 and t2, with a time interval Δt = t2-t1, the position coordinates of the particle in three-dimensional space are obtained by the position monitoring sensor as (x1, y1, z1) and (x2, y2, z2), and the displacement of the particle in the time Δt is calculated. The particle flow rate Among them, (x1, y1, z1) and (x2, y2, z2) are the spatial position coordinates of the particles at different times. The flow velocity is obtained by the ratio of displacement to time, which intuitively reflects the speed of particle movement in the pipeline.

4. A lubrication equipment component failure prediction test system according to claim 1, characterized in that: The particle feature solution module calculates the particle concentration as follows: the lubrication pipeline is divided into n small sections along the length direction, and any lubrication pipeline area with a length of L is selected from the lubrication pipeline. The cross-sectional area of the area is S. The number of particles in the area detected by the multi-frequency eddy current sensor group is N. The volume of the area is V = S × L, and the particle concentration is That is, the number of particles per unit volume, which is used to measure the density of particles in the pipeline. The material, position, flow rate, and concentration of the particles are used as the particle data sequence D, and a real-time concentration heat map of the material distribution is generated.

5. A lubrication equipment component failure prediction test system according to claim 1, characterized in that: The entropy value analysis module calculates the entropy change value of the microelement based on the real-time temperature and pressure data of the multi-source sensing module, that is, Among them, m is the number of temperature sensors, ΔS is the entropy change value of the microelement, q i is the heat flux density between the i-th adjacent temperature measurement points and k is thermal conductivity, T i+1 and T i are the temperature values of two adjacent temperature measuring points, L is the length between the adjacent temperature measuring points, A is the heat transfer area, P is the pressure, ΔV gas is the change in gas volume.

6. A lubrication equipment component failure prediction test system according to claim 5, characterized in that: The process of threshold matching performed by the entropy analysis module is as follows: When the real-time entropy change ΔS meets the conditions, the phase change flag F is triggered. phase ∈{0,1,2}, that is, when |ΔS-S SL |<0.05S SL When the phase change flag F is set phase =1 indicates solid-liquid phase transition; When |ΔS-S LG |<0.05S LG When the phase change flag F is set phase =2 represents liquid-gas phase transition, where S SL is the entropy of the solid-liquid phase transition of the lubricant, S LG is the entropy of the lubricant liquid-gas phase transition, F phase =0 means no phase change, which is used to judge the phase change state of the lubricant and output the results to the collaborative diagnosis module, including: entropy change value ΔS, phase change flag F phase and time t, that is, {ΔS,F phase ,t}.

7. A lubrication equipment component failure prediction test system according to claim 1, characterized in that: The collaborative diagnosis module uses dynamic time warping to align particle data with entropy change time series and finds the best match between the two sequences by bending the time axis, i.e. Where DT(D, ΔS) represents the dynamic time warping distance between the particle data sequence D and the entropy change data sequence ΔS, π represents the set of all time warping paths, (i, j) is the point on the path π, corresponding to the element in the i-th region of the particle data sequence D and the j-th element in the entropy change data sequence ΔS, D(i) and ΔS(j) are the values of the particle data sequence and the entropy change data sequence at the corresponding positions, respectively, μ D and μ S are the means of the particle data sequence and the entropy change data sequence, σ D and σ S They are the standard deviations of the particle data sequence and the entropy change data sequence respectively. The optimal time warping path is found through dynamic time warping to align the time and space of the particle data and the entropy change data.

8. A lubrication equipment component failure prediction test system according to claim 7, characterized in that: The collaborative diagnosis module establishes an association rule base, determines the type of equipment fault by setting different conditions, and calculates the confidence level of each rule. The fault types include bearing overheating and wear and lubricant decomposition and cavitation. The rules of the association rule base are: Bearing overheating wear; when the metal particle concentration C js More than 15% and lasts for 5 minutes, and the phase change mark F phase =1, trigger the rule and calculate the confidence α1 = 0.7·sigmoid(C js / 15%)+0.3·sigmoid(ΔS / S SL ), where α1 is the confidence level of bearing overheat wear, It is an activation function that maps the input value to between 0 and 1 and quantifies the impact of different factors on the confidence level. x represents the input value, C js is the metal particle concentration, ΔS is the entropy change value, S SL is the entropy of the solid-liquid phase transition of the lubricant; Lubricant decomposition cavitation: When the non-metallic particle concentration C fjs More than 300%, and the phase change mark F phase =2, the rule is triggered and the confidence α2 = 0.5·tanh(C fjs / 300%)+0.5·exp(-ΔS / S LG ), where α2 is the lubricant decomposition cavitation confidence, tanh(x) is the hyperbolic tangent function, which is used to map the input value to between -1 and 1 to quantify the influence of non-metallic particle concentration, exp(x) is the exponential function, which is used to describe the influence of entropy change on confidence, C fjs is the concentration of non-metallic particles, S LG is the entropy of the lubricant liquid-gas phase transition; When two rules are triggered at the same time, the maximum confidence is taken as the final confidence α final , and output the corresponding fault type F fault ∈{1,2}, that is, α final =max(α1,α2), where α final is the final confidence level, when α final When α1 is maximum, F fault =1, indicating bearing overheating and wear failure, when α final When α2 is maximum, F fault =2, indicating lubricant decomposition cavitation failure.

Images