CN113516273B

CN113516273B - Diesel engine supercharger fault prediction method for power generation

Info

Publication number: CN113516273B
Application number: CN202110369737.0A
Authority: CN
Inventors: 杨天诣; 程垠钟; 杜剑维
Original assignee: Military Products Technology Research Center Of China Shipbuilding Industry Corp
Current assignee: Military Products Technology Research Center Of China Shipbuilding Industry Corp
Filing date: 2021-04-02
Publication date: 2024-06-04
Anticipated expiration: 2041-04-02

Abstract

The invention discloses a failure prediction method of a diesel supercharger for power generation, which comprises the following steps of: collecting unit information to form a data set: step 2: processing data, and primarily judging fault data characteristics: step 3: according to the working characteristics of Chai Fa units running in parallel, the noise reduction method is improved, and fault characteristic analysis and threshold determination are respectively carried out under the power factors of 0.6, 0.8 and 0.9 according to the load working conditions of 20%, 50%, 75%, 90%, 75%, 50% and 20%: step 4: summarizing and analyzing, improving a threshold determining method, and obtaining thresholds of different working conditions: step 5: forming a multi-threshold expression, and defining the result of judging the abnormal situation through the multi-threshold expression to finish off-line processing: step 6: and the prediction is realized on line. According to the diesel engine supercharger fault prediction method for the power generation, the relation between the supercharger lubricating oil inlet pressure and the active power, the supercharger lubricating oil inlet temperature and the diesel engine lubricating oil pressure is read, and the prediction accuracy is improved through setting of a plurality of same-level thresholds.

Description

Diesel engine supercharger fault prediction method for power generation

Technical Field

The invention relates to the technical field of diesel engine fault prediction, in particular to a diesel engine supercharger fault prediction method for power generation.

Background

In a land joint debugging test of a large ship electric power system, a plurality of diesel generator sets run in parallel for a long time, the parallel running stability of the diesel generator sets is checked, and the main running working conditions of the diesel generator sets are as follows according to relevant standards: parallel operation tests of loading and unloading were performed at 20%, 50%, 75%, 90%, 75%, 50%, 20% load conditions at power factors of 0.6, 0.8 and 0.9.

The supercharger of the diesel engine for power generation works by using the exhaust gas of the diesel engine and the surplus energy after the exhaust gas through the turbine, so that the compressor can provide more compressed air for the diesel engine to achieve the optimal operation performance. Because the air entering the cylinder is increased, the fuel oil can be combusted more fully, so that the diesel engine generates more power, reduces emission and pollution. The main categories of supercharger faults include assembly quality problems of a supercharger body, unqualified part manufacturing, overlarge rotor unbalance, bearing design and machining problems, overspeed running of a unit, entering of external foreign matters into the supercharger, oil way blockage of a lubricating system, insufficient oil supply of the lubricating system and the like, wherein bearing failure caused by lubrication is a common fault with higher probability.

Disclosure of Invention

The invention provides a fault prediction method of a diesel engine supercharger for power generation, which aims to solve the problem that in the prior art, the detection method of the lubrication abnormality of the supercharger is to set a fixed threshold value for the too low oil inlet pressure of the supercharger, and send an alarm signal when the actual pressure value is lower than the threshold value. However, in the actual process, when the alarm signal is received, the supercharger has failed, and the prediction and prevention of the failure cannot be realized. In the actual working process, the oil inlet pressure values of the lubricating oil of the supercharger are different under different working conditions, and a single threshold value cannot be used for carrying out real-time fault identification and trend judgment on the oil inlet pressure of the lubricating oil which is too low.

In order to solve the technical problems, the invention adopts the following technical scheme:

a method for predicting faults of a diesel supercharger for power generation comprises the following steps:

step 1, collecting unit information to form a data set;

collecting all relevant operation data of the fault diesel engine unit and the normal diesel engine unit from the beginning of the input test to the fault occurrence time, and collecting the operation data after the fault diesel engine unit is repaired;

Step 2, processing data, primarily judging fault data characteristics, judging useful information and parameter correlation so as to determine the input and output of multi-threshold expression judgment;

Step 3, improving a noise reduction method according to the working characteristics of the parallel operation of diesel-electric sets of the land joint debugging test of the large-scale ship electric power system, and respectively carrying out fault characteristic analysis and threshold determination according to the load working conditions of 20%, 50%, 75%, 90%, 75%, 50% and 20% under the power factors of 0.6, 0.8 and 0.9;

step 4, summarizing and analyzing, and improving a threshold determining method to obtain thresholds of different working conditions;

Step 5, forming a multi-threshold expression, and prescribing a result of judging the abnormal condition through the multi-threshold expression to finish off-line processing;

And 6, on-line prediction is realized.

Further, in step 2, the method further comprises the following steps:

step 2.1, noise reduction processing is carried out on the collected unit information: firstly, screening and removing noise collected by a sensor at the moment of starting a diesel engine, then carrying out noise reduction treatment by using a moving average filter, primarily judging the numerical value and the variation trend of the oil inlet pressure of the supercharger in the same working condition and different working conditions of a fault diesel engine unit and other units, primarily judging the characteristics of fault data, and primarily judging parameters influencing the variation of the oil inlet pressure of the supercharger;

Step 2.2, input and output of the correlation judgment determination threshold judgment: determining that the active power, the supercharger oil inlet temperature and the diesel engine oil pressure are related to the change of the supercharger oil inlet pressure, performing correlation judgment on the parameters as useful information, namely respectively calculating a Pirson correlation coefficient and a Speman correlation coefficient, finding that the ratio of the supercharger oil inlet pressure to the diesel engine oil pressure is respectively and inversely related to the active power and the supercharger oil inlet temperature, the active power and the supercharger oil inlet temperature are highly and positively related, determining that the threshold input is an arrangement combination of the active power and the supercharger oil inlet temperature, and the threshold output is respectively the ratio of the supercharger oil inlet pressure to the diesel engine oil pressure.

Further, in step 3, the method further comprises the following steps:

Step 3.1, determining a subsequent analysis mode through the working characteristics of the parallel operation of the diesel generator sets of the land joint debugging test of the large-scale ship power system:

In the parallel operation process of the diesel generator set of the land joint debugging test of the large ship electric power system, parallel operation tests of loading and unloading are respectively carried out under the power factors of 0.6, 0.8 and 0.9 according to the load working conditions of 20%, 50%, 75%, 90%, 75%, 50% and 20%, the active power of different power factors is different under the same load working conditions, and the corresponding booster lubricating oil inlet pressure, the ratio of the booster lubricating oil inlet pressure to the diesel lubricating oil pressure is different, so that specific analysis is carried out according to different power factors;

Step 3.2, improving a noise reduction method, and respectively carrying out fault characteristic analysis according to 20%, 50%, 75% and 90% of load working conditions at power factors of 0.6, 0.8 and 0.9:

Step 3.2.1, improving a noise reduction method, and classifying data into noise reduction according to the power factors of 0.6, 0.8 and 0.9, and the load conditions of 20%, 50%, 75%, 90%, 75%, 50% and 20%: extracting data in each test process from the collected data set through a programming language, classifying according to 0.6, 0.8 and 0.9 power factor operation conditions, extracting the oil inlet pressure of the supercharger under the conditions of 20%, 50%, 75% and 90% of load and other related information from each type of data, and carrying out averaging treatment on the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine and the active power batch;

step 3.2.2, determining a fault characteristic analysis method:

when fault feature judgment is carried out, respectively comparing and judging the data values corresponding to the same names in the fault and normal data sets to be:

a. The height under the same working condition;

b. multiple test change trends of the same working condition;

c. Different working condition change conditions are tested at the same time;

d. c, a change trend in the process of multiple tests;

Wherein a and b can be directly judged by improving the data set after the noise reduction treatment, and d can be directly judged by c obtained by a plurality of tests after c is obtained; in the step 2.2, the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are determined to be respectively and highly inversely related to the active power, and the invention carries out linear fitting on the extracted oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine and the average value of the active power in each experimental process under different power factors respectively, and is used for judging c, wherein the expression of a fitted curve is as follows:

y＝ax+b (1)

wherein a is a slope, and represents the variation trend of the oil inlet pressure and the ratio of the lubricating oil of the supercharger in the test process; b is the intercept, and represents the integral size of the oil inlet pressure and the ratio of the lubricating oil of the supercharger; x is active power; y is the oil inlet pressure of the lubricating oil of the supercharger;

Step 3.2.3, performing fault characteristic analysis:

Firstly judging the level of data values corresponding to the same name in fault and normal data sets under the same working condition, fitting a and b obtained by fitting curves according to time sequences to obtain a fitting scatter diagram and a fitting line diagram and analyze trends, wherein the multiple sets processed in the step 4 are respectively subjected to 0.6, 0.8, 0.9 power factors, 20%, 50%, 75% and 90% load conditions;

In the analysis process, firstly solving the variance of the data under different power factors, and judging the convergence of the data; and performing quadratic polynomial fitting on the obtained fitting scatter diagram and fitting line diagram, wherein the expression of the fitting curve is as follows:

y＝a₀t²+b₀t+c₀ (2)

wherein: y is the oil inlet pressure of the lubricating oil of the supercharger; t is the test times; deriving the expression, obtaining the maximum value of the pole position t ₀,t₀ of the fitted curve in the test of the number of times, judging the descending trend by matching with the quadratic term coefficient, and combining the times of the test under different conditions together with n; the specific judgment criteria are as follows:

If a ₀＞0,t₀ is less than or equal to 0, b obtained by a ratio of the oil inlet pressure of the supercharger oil to the oil inlet pressure of the diesel engine/fitting a curve is in an ascending trend in the process of multiple tests;

If a ₀＞0,0＜t₀ is less than or equal to n, the oil inlet pressure of the supercharger oil/the ratio of the oil inlet pressure of the supercharger oil to the oil pressure of the diesel engine/b obtained by fitting a curve are firstly reduced and then increased in the process of multiple tests;

if a ₀＞0,t₀ is more than n, b obtained by a ratio of the oil inlet pressure of the supercharger oil to the oil pressure of the diesel engine/fitting a curve shows a descending trend in the process of multiple tests;

If a ₀＜0,t₀ is less than or equal to 0, b obtained by a ratio of the oil inlet pressure of the supercharger oil to the oil inlet pressure of the diesel engine/a fitting curve shows a descending trend in the process of multiple tests;

if a ₀＜0,0＜t₀ is less than or equal to n, the oil inlet pressure of the supercharger oil/the ratio of the oil inlet pressure of the supercharger oil to the oil pressure of the diesel engine/b obtained by fitting a curve are firstly increased and then decreased in the process of multiple tests;

If a ₀＜0,t₀ is more than n, b obtained by a ratio of the oil inlet pressure of the supercharger oil to the oil pressure of the diesel engine/fitting a curve is in an ascending trend in the process of multiple tests;

Compared with a normal unit, the failure unit has the advantages that the oil inlet pressure of the supercharger of the failure unit, the ratio variance of the oil inlet pressure of the supercharger and the oil pressure of the diesel engine is large and is in a descending trend, the overall numerical value is smaller, and the variation trend of the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger and the oil pressure of the diesel engine is unchanged under different working conditions in each experimental process, namely a is unchanged; finally, determining that the fault characteristics are that the oil inlet pressure of the supercharger lubricating oil and the ratio value of the oil inlet pressure of the supercharger lubricating oil to the lubricating oil pressure of the diesel engine are smaller, and the overall trend is shown;

Step 3.3, determining a threshold value by using the box diagram:

And 3.3.1, wherein the fault is characterized in that the values of the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are smaller and continuously reduced, so that the values of the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are respectively set to be low threshold values, and the box type diagram is used for respectively solving two threshold values of the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine under the conditions of 0.6, 0.8, 0.9 power factor, 20%, 50%, 75% and 90% of load, wherein the threshold value with larger value is defined as a pre-alarm value, and the threshold value with lower value is defined as an alarm value.

Further, in step 4, the method further comprises the following steps:

step 4.1, summarizing thresholds under different load conditions of 0.6, 0.8 and 0.9 power factors to form a threshold table, and specifically judging abnormal conditions under different load conditions of 0.6, 0.8 and 0.9 power factors by comparing the relation between actual values and values in the threshold table in the later period;

And 4.2, screening out a part of data with more operation times according to different active powers by using a programming language, respectively carrying out averaging treatment on the supercharger lubricating oil inlet pressure, the ratio of the supercharger lubricating oil inlet pressure to the diesel engine lubricating oil pressure under the same active power and different times of tests, and respectively solving out two thresholds of the ratio of the supercharger lubricating oil inlet pressure, the supercharger lubricating oil inlet pressure to the diesel engine lubricating oil pressure by using a box graph, wherein the threshold with larger value is defined as a pre-alarm value, and the threshold with smaller value is defined as an alarm value.

Further, in step 5, the method further comprises the following steps:

Step 5.1, forming a multi-threshold expression: step 2.2 shows that the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are respectively and highly inversely related to the active power, so that thresholds under different active powers are summarized together and fitted into corresponding curves, and the maximum times of polynomials of the curves are determined by comparing residual errors, namely two threshold expressions of the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are respectively obtained;

step 5.2, the step 2.2 shows that the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil inlet pressure of the diesel engine are respectively and highly inversely related to the active power and the oil inlet temperature of the supercharger, and the active power and the oil inlet temperature of the supercharger are highly positively related, so that the oil inlet pressure of the supercharger can be commonly represented by the active power and the oil inlet temperature of the supercharger in the actual process, and the method is more rigorous than judging by only the threshold expression of the oil inlet pressure of the supercharger and the active power;

Step 5.2.1, when a large number of samples are possessed, classifying the data screened in the step 4.2 again according to the oil inlet temperature of the supercharger, respectively carrying out averaging treatment on the same active power, the same oil inlet temperature of the supercharger, the oil inlet pressure of the supercharger under different tests and the ratio of the oil inlet pressure of the supercharger to the oil inlet pressure of the diesel engine, respectively solving two thresholds by using a box graph, gathering the thresholds under different active powers and the oil inlet temperatures of different superchargers together, and fitting the thresholds into corresponding curves, namely obtaining two oil inlet pressure threshold expressions of the supercharger, which are jointly determined by the active power and the oil inlet temperature of the supercharger under the condition that the samples are sufficient;

Step 5.2.2, when the number of samples is insufficient to support the treatment of step 5.2.3, performing multiple fitting on the obtained data in each test process, wherein the fitted curve expression is:

P＝a₂T+b₂P_w+c₂ (3)

Wherein P is the oil inlet pressure of the lubricating oil of the supercharger; p _w is active power; t is the lubricating oil inlet temperature of the supercharger;

Setting the same and proper oil inlet temperature and active power of the supercharger lubricating oil for each test, and carrying out threshold selection on the calculated oil inlet pressure of the supercharger lubricating oil by utilizing a box graph, wherein the corresponding expressions are two threshold expressions;

A total of 6 threshold expressions are finally determined, in the form of:

y₁＝a₁xⁿ+b₁xⁿww+… (4)

y₂＝a₁xⁿ+b₁x^n-1+… (5)

y₃＝a₂xⁿ+b₂x^n-1+… (6)

y₄＝a₂xⁿ+b₂x^n-1+… (7)

y₅＝a₁T+b₁P_W+c₁ (8)

y₆＝a₂T+b₂P_W+c₂ (9)

Wherein x is the unit active power, namely the ratio of the active power to the rated power; p _w is active power; t is the lubricating oil inlet temperature of the supercharger; y ₁ represents the supercharger oil inlet pressure, y ₂ represents the ratio of the supercharger oil inlet pressure to the diesel engine oil pressure, y ₃ represents the supercharger oil inlet pressure, y ₄ represents the ratio of the supercharger oil inlet pressure to the diesel engine oil pressure, and y ₅、y₆ represents the supercharger oil inlet pressure;

finally, three pre-alarm expressions are denoted by y ₁、y₂、y₅ and three alarm expressions are denoted by y ₃、y₄、y₆;

step 5.3, determining the degree of abnormality:

Let W be the current supercharger oil inlet pressure or the ratio of supercharger oil inlet pressure to diesel engine oil pressure, W ₁ and W ₂ be the pre-alarm value and the alarm value under the current working condition:

When W ₁ is less than W, the supercharger is lubricated normally;

When W ₂＜W≤W₁ is carried out, the oil pre-alarming of the supercharger is carried out;

And when W is less than or equal to W ₂, the oil lubrication of the supercharger is alarmed.

Further, in step 6, the method further comprises the following steps:

Step 6.1, inputting the oil inlet temperature P of the supercharger lubricating oil, the oil inlet temperature T of the supercharger lubricating oil, the active power P _w and the lubricating oil pressure PD of the diesel engine in real time through a data bus, preprocessing by utilizing a multi-threshold expression, outputting W, W ₁ and W ₂ under the current working condition, and entering a model to judge abnormal conditions;

step 6.2, outputting a prediction result:

When the pre-alarm is continuously generated, the oil inlet pressure of the lubricating oil of the supercharger is lower and the lubricating oil is in a descending trend, impurities begin to exist in the lubricating oil, corresponding measures are recommended to be taken to increase the pressure, and corresponding inspection is carried out;

When an alarm occurs, the existence of the interval range of the two threshold values indicates that the oil inlet pressure of the lubricating oil of the supercharger is seriously low, the pressure drop is rapid and obvious, the supercharger is in a high probability of failure, the machine is stopped immediately for checking, and corresponding measures are taken to increase the pressure.

Compared with the prior art, the fault prediction method for the diesel supercharger for power generation has the following remarkable and superior effects:

The invention provides a fault prediction method of a diesel supercharger for power generation, which can be used for carrying out real-time abnormal information identification and fault prediction under different working conditions after a multi-threshold expression is determined, and judging the degree of abnormal conditions.

The invention provides a failure prediction method of a diesel engine supercharger for power generation, which is characterized in that the relation between the oil inlet pressure and active power of the supercharger lubricating oil, the oil inlet temperature of the supercharger lubricating oil and the oil inlet pressure of the diesel engine is read, and the prediction precision is increased through setting up a plurality of same-level thresholds.

Drawings

Fig. 1 is a schematic flow chart of a method for predicting faults of a diesel supercharger for power generation.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is evident that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by a person skilled in the art without any inventive effort, are intended to be within the scope of the present invention, based on the embodiments of the present invention.

As shown in fig. 1, the embodiment of the invention provides a method for predicting a failure of a diesel supercharger for power generation, which comprises the following steps:

step 1, collecting unit information to form a data set;

And 6, on-line prediction is realized.

Further, in step 2, the method further comprises the following steps:

Further, in step 3, the method further comprises the following steps:

step 3.2.2, determining a fault characteristic analysis method:

a. The height under the same working condition;

b. multiple test change trends of the same working condition;

c. Different working condition change conditions are tested at the same time;

d. c, a change trend in the process of multiple tests;

y＝ax+b (1)

Step 3.2.3, performing fault characteristic analysis:

In the analysis process, firstly solving the variance of the data under different power factors, and judging the convergence of the data: and performing quadratic polynomial fitting on the obtained fitting scatter diagram and fitting line diagram, wherein the expression of the fitting curve is as follows:

y＝a₀t²+b₀t+c₀ (2)

Wherein: y is the oil inlet pressure of the lubricating oil of the supercharger; t is the test times; deriving the expression, obtaining the pole position t ₀ of the fitting curve, and judging the descending trend by matching with the quadratic term coefficient, wherein the times of the test under different conditions are collectively called n; the specific judgment criteria are as follows:

Step 3.3, determining a threshold value by using the box diagram:

Further, in step 4, the method further comprises the following steps:

Further, in step 5, the method further comprises the following steps:

P＝a₂T+b₂P_w+c₂ (3)

A total of 6 threshold expressions are finally determined, in the form of:

y₁＝a₁xⁿ+b₁x^n-1+… (4)

y₂＝a₁xⁿ+b₁x^n-1+… (5)

y₃＝a₂xⁿ+b₂x^n-1+… (6)

y₄＝a₂xⁿ+b₂x^n-1+… (7)

y₅＝a₁T+b₁P_W+c₁ (8)

y₆＝a₂T+b₂P_W+c₂ (9)

step 5.3, determining the degree of abnormality:

When W ₁ is less than W, the supercharger is lubricated normally;

Further, in step 6, the method further comprises the following steps:

step 6.2, outputting a prediction result:

The invention relates to a fault prediction method of a diesel supercharger for power generation, which comprises the following steps:

Step 1: collecting unit information to form a data set: collecting all relevant operation data of the fault diesel engine unit and the normal diesel engine unit from the beginning of the input test to the fault occurrence time, and collecting the operation data after the fault diesel engine unit is repaired;

Step 2: processing data, primarily judging fault data characteristics, judging useful information and parameter correlation so as to determine input and output of multi-threshold expression judgment:

Step 3: according to the working characteristics of the parallel operation of diesel-electric sets of the land joint debugging test of the large-scale ship electric power system, the noise reduction method is improved, and fault characteristic analysis and threshold determination are respectively carried out under the power factors of 0.6, 0.8 and 0.9 according to the load working conditions of 20%, 50%, 75%, 90%, 75%, 50% and 20%:

step 4: summarizing and analyzing, improving a threshold determining method, and obtaining thresholds of different working conditions:

step 5: forming a multi-threshold expression, and defining the result of judging the abnormal situation through the multi-threshold expression to finish off-line processing:

Step 6: on-line implementation prediction:

The step2 specifically comprises the following steps:

Step 2.1: noise reduction processing is carried out on the collected unit information: firstly, screening and removing noise collected by a sensor at the moment of starting a machine, then carrying out noise reduction treatment by using a moving average filter, primarily judging the numerical value and the variation trend of the oil inlet pressure of the supercharger lubricating oil under the same working condition and different working conditions of a fault unit and other units, primarily judging fault data characteristics, and primarily judging parameters influencing the variation of the oil inlet pressure of the supercharger lubricating oil;

Step 2.2: input/output of correlation judgment determination threshold judgment: determining that the active power, the oil inlet temperature of the supercharger lubricating oil and the oil inlet pressure of the diesel engine are related to the change of the oil inlet pressure of the supercharger lubricating oil, and carrying out correlation judgment on the parameters as useful information, namely respectively calculating a Pearson (Pearson) correlation coefficient and a Spearman (Spearman) correlation coefficient

Table 1 parameter correlation coefficient table

When the absolute value of each parameter is more than or equal to 0.7, the two parameters can be called as being highly (linearly) related, and then the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil inlet pressure of the diesel engine can be found to be respectively and inversely related to the active power and the oil inlet temperature of the supercharger, and the active power and the oil inlet temperature of the supercharger are positively related; thus, the threshold input is determined to be the arrangement combination of the active power and the oil inlet temperature of the supercharger, and the threshold output is respectively the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine.

The step 3 specifically comprises the following steps:

step 3.1: the method for carrying out subsequent analysis is determined through the working characteristics of the parallel operation of the diesel generator sets of the land joint debugging test of the large-scale ship power system: in the parallel operation process of the diesel generator set of the land joint debugging test of the large ship electric power system, parallel operation tests of loading and unloading are respectively carried out under the power factors of 0.6, 0.8 and 0.9 according to the load working conditions of 20%, 50%, 75%, 90%, 75%, 50% and 20%, the active power of different power factors is different under the same load working conditions, and the corresponding ratio of the oil inlet pressure of the supercharger lubricating oil to the oil inlet pressure of the diesel engine is different, so that the invention carries out specific analysis according to different power factors.

Step 3.2: the improved noise reduction method is characterized in that fault characteristic analysis is carried out under the power factors of 0.6, 0.8 and 0.9 according to the load working conditions of 20%, 50%, 75% and 90% respectively:

Step 3.2.1: the noise reduction method is improved, and the data are subjected to noise reduction classification according to the load conditions of 20%, 50%, 75%, 90%, 75%, 50% and 20% with the power factors of 0.6, 0.8 and 0.9: the noise reduction result in the step 2.1 contains the oil inlet pressure value of the oil of the supercharger under various working conditions, the accurate judgment cannot be carried out, and when a windowing noise reduction method is adopted, the pressure values under the conditions of multiple tests, different power factors and different loads are fused together, so the improvement is needed as follows. The data in each test process are extracted from the collected data set through a programming language, the data are classified according to the 0.6, 0.8 and 0.9 power factor operation working conditions, the oil inlet pressure of the supercharger under the conditions of 20%, 50%, 75% and 90% of load and other relevant information are extracted from each type of data, the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine and the active power are respectively subjected to average treatment in batches, and by the method, the pressure values under different tests, different power factors and different loads can be separated, meanwhile, the noise reduction effect of a sliding average filter can be achieved, and accordingly corresponding judgment and subsequent calculation can be more intuitively carried out.

Step 3.2.2: determining a fault characteristic analysis method:

a. The height under the same working condition is equal to that of the water,

B. The same working condition is subjected to multiple test on the change trend,

C. different working condition change conditions are tested at the same time,

D. c trend of change over the course of multiple trials,

Wherein a and b can be directly judged by improving the data set after the noise reduction treatment, and d can be directly judged by c obtained by multiple experiments after c is obtained. In the step 2.2, the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are determined to be respectively and highly inversely related to the active power, and the invention carries out linear fitting on the extracted oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine and the average value of the active power in each experimental process under different power factors respectively, and is used for judging c, wherein the expression of a fitted curve is as follows:

y＝ax+b (1)

a is a slope, and represents the variation trend of the oil inlet pressure and the ratio of the lubricating oil of the supercharger in the test process;

b is the intercept, and represents the integral size of the oil inlet pressure and the ratio of the lubricating oil of the supercharger;

x is active power;

y is the oil inlet pressure of the lubricating oil of the supercharger.

Step 3.2.3: and (3) performing fault characteristic analysis:

Firstly judging the level of data values corresponding to the same name in fault and normal data sets under the same working condition, then fitting the supercharger lubricating oil inlet pressure, the ratio of the supercharger lubricating oil inlet pressure to the diesel lubricating oil pressure and the fitting curve obtained by a and b under the conditions of 0.6, 0.8, 0.9 power factors, 20%, 50%, 75% and 90% loads of a plurality of units treated in the step 4 according to the time sequence (test times), obtaining a fitting scatter diagram and a fitting line diagram, and analyzing trend.

In the analysis process, firstly solving the variance of the data under different power factors, and judging the convergence of the data; performing a second polynomial fitting on the obtained fitting scatter diagram and fitting line diagram, deriving the obtained expression, obtaining the pole position t ₀ of the fitting curve, and judging the descending trend by matching with the quadratic term coefficient, wherein 1 and 2 are descending, and 0 is the other;

Table 2 multiple test trends of oil inlet pressure for supercharger lubricating oil

Compared with a normal unit, the failure unit has large (divergent) ratio variance of booster lubricating oil inlet pressure and booster lubricating oil inlet pressure to diesel engine lubricating oil pressure, and has a descending trend, and the overall numerical value is smaller. And finally, determining that the fault characteristics are that the oil inlet pressure of the supercharger lubricating oil and the ratio value of the oil inlet pressure of the supercharger lubricating oil to the lubricating oil pressure of the diesel engine are smaller, and the overall trend is shown.

Step 3.3: determining a threshold value by using a box graph:

step 3.3.1: the fault characteristic is that the oil inlet pressure of the supercharger lubricating oil and the ratio value of the oil inlet pressure of the supercharger lubricating oil to the oil pressure of the diesel engine are smaller and continuously reduced, so that the oil inlet pressure and the oil inlet pressure of the supercharger lubricating oil and the oil inlet pressure of the diesel engine are respectively set with low threshold values;

The box graph is a statistical graph used for displaying a group of data dispersion situation data; the box graph has the greatest advantages that the box graph is not influenced by abnormal values, and the discrete distribution condition of data can be accurately and stably depicted; it can display the upper edge value, lower edge value, median, upper quartile (Q3), lower quartile (Q1) and outlier of a set of data. The quartile is to arrange all values from small to large and divide them into four equal divisions, the value at the position of three division points is the quartile, the quartile distance iqr=q3-Q1, the value at q3+1.5iqr and q1-1.5iqr is located at the abnormal value cutoff point, called the inner limit (upper and lower edge values), the value at q3+3iqr and q1-3iqr is called the outer limit, the value outside the inner limit is the abnormal value, the value between the inner limit and the outer limit is the mild abnormal value, and the value outside the outer limit is the extreme abnormal value. Because the box diagram has the function of eliminating abnormal values, the integral booster oil inlet pressure of the fault unit, the ratio of the booster oil inlet pressure to the diesel engine oil inlet pressure are far smaller than those of other normal units, the box diagram has a descending trend, and in order to leave a certain margin, the upper quartile and the lower quartile of the box diagram are selected as two thresholds under each condition;

and respectively solving two thresholds of the supercharger lubricating oil inlet pressure and the ratio of the supercharger lubricating oil inlet pressure to the diesel engine lubricating oil pressure under the conditions of 0.6, 0.8, 0.9 power factors and 20%, 50%, 75% and 90% loads by utilizing a box graph, wherein the threshold with a larger value is defined as a pre-alarm value, and the threshold with a lower value is defined as an alarm value.

The step 4 specifically comprises the following steps:

Step 4.1: summarizing thresholds of the power factors of 0.6, 0.8 and 0.9 under different load conditions to form a threshold table, and judging abnormal conditions of the power factors of 0.6, 0.8 and 0.9 under different load conditions by comparing the relation between actual values and values in the threshold table in the later period;

step 4.2: the method comprises the steps of screening fault unit data according to different active powers by using a programming language, screening out a part of data with more operation times, respectively carrying out averaging treatment on the oil inlet pressure of the supercharger lubricating oil and the ratio of the oil inlet pressure of the supercharger lubricating oil to the oil pressure of the diesel engine under the same active power and different times, respectively solving two thresholds of the oil inlet pressure of the supercharger lubricating oil and the ratio of the oil inlet pressure of the supercharger lubricating oil to the oil pressure of the diesel engine by using a box graph, wherein the threshold with larger value is defined as a pre-alarm value, and the threshold with lower value is defined as an alarm value;

The step 5 specifically comprises the following steps:

Step 5.1: forming a multi-threshold expression: as can be seen from step 2.2, the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are respectively and highly inversely related to the active power, so that the thresholds under different active powers can be summarized together and fitted into corresponding curves, and the maximum degree of the polynomial of the curves can be determined by comparing the residual errors, so that two threshold expressions of the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine can be respectively obtained.

Step 5.2: performing multiple fitting on the obtained data in each test process, wherein the fitted curve expression is as follows:

P＝a₂T+b₂P_w+c₂ (3)

Wherein: p is the oil inlet pressure of the lubricating oil of the supercharger;

p _w is active power;

T is the lubricating oil inlet temperature of the supercharger;

A total of 6 threshold expressions are finally determined, in the form of:

y₁＝a₁x²+b₁x+c₁ (4)

y₂＝a₁x²+b₁x+c₁ (5)

y₃＝a₂x²+b₂x+c₂ (6)

y₄＝a₂x²+b₂x+c₂ (7)

y₅＝a₁T+b₁P_W+c₁ (8)

y₆＝a₂T+b₂P_W+c₂ (9)

wherein: x is the unit active power, namely the ratio of the active power to the rated power;

p _w is active power;

T is the lubricating oil inlet temperature of the supercharger;

y ₁、y₃ represents the supercharger oil feed pressure, the ratio of y ₂、y₄ supercharger oil feed pressure to diesel engine oil pressure, and y ₅、y₆ represents the supercharger oil feed pressure.

after comparing and analyzing the threshold curves with other unit parameters, the ratio of the oil inlet pressure of the supercharger lubricating oil of the normal unit and the repaired unit/the oil inlet pressure of the supercharger lubricating oil of the supercharger to the oil inlet pressure of the diesel engine is generally larger than a corresponding first threshold, only one unit is slightly smaller than a pre-alarm value under the extremely individual condition, the oil inlet pressure of the supercharger lubricating oil of the unit is generally smaller than that of other normal units, and the diameter of an oil inlet hole plate is enlarged due to the lower oil inlet pressure of the supercharger in the later period; the ratio of the oil inlet pressure of the supercharger lubricating oil of all the units/the oil inlet pressure of the supercharger lubricating oil to the oil pressure of the diesel engine is larger than a corresponding second threshold value, which indicates that the threshold value is reasonable to set;

Step 5.3: determining the degree of abnormality:

Let W be the current oil inlet pressure of the supercharger or the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine, W ₁ and W ₂ be the pre-alarm value and the alarm value under the current working condition,

When W ₁ is less than W, the supercharger is lubricated normally;

when W is less than or equal to W ₂, the oil lubrication of the supercharger is alarmed;

the step 6 specifically comprises the following steps:

Step 6.1: inputting P (supercharger lubricating oil inlet temperature), T (supercharger lubricating oil inlet temperature), P _w (active power) and PD (diesel engine lubricating oil pressure) in real time through a data bus, preprocessing by utilizing a multi-threshold expression, outputting W, W ₁ and W ₂ under the current working condition, and entering a model to judge abnormal conditions;

step 6.2: outputting a prediction result:

When the pre-alarm is continuously generated, the oil inlet pressure of the lubricating oil of the supercharger is lower and is in a descending trend, impurities possibly exist in the lubricating oil, corresponding measures are recommended to be taken to increase the pressure, and corresponding inspection is carried out;

When an alarm occurs, as a certain interval range exists between the two threshold values, the pressure of the lubricating oil inlet of the supercharger is seriously lower, the pressure drop is rapid and obvious, the supercharger is in a high probability of failure, the machine is stopped immediately for checking, and corresponding measures are taken to increase the pressure.

In the description of the present specification, a particular feature, structure, material, or characteristic may be combined in any suitable manner in one or more embodiments or examples.

Of course, the present invention is capable of other various embodiments and its several details are capable of modification and variation in light of the present invention by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. The fault prediction method for the diesel supercharger for power generation is characterized by comprising the following steps of:

step 1, collecting unit information to form a data set;

Step 2.2, input and output of the correlation judgment determination threshold judgment: determining that the active power, the supercharger oil inlet temperature and the diesel engine oil pressure are related to the change of the supercharger oil inlet pressure, performing correlation judgment on the parameters as useful information, namely respectively calculating a Pirson correlation coefficient and a Speman correlation coefficient, finding that the ratio of the supercharger oil inlet pressure to the diesel engine oil pressure is respectively and inversely related to the active power and the supercharger oil inlet temperature, the active power and the supercharger oil inlet temperature are highly and positively related, determining that the threshold input is an arrangement combination of the active power and the supercharger oil inlet temperature, and the threshold output is respectively the ratio of the supercharger oil inlet pressure to the diesel engine oil pressure;

step 3.2.2, determining a fault characteristic analysis method:

a. The height under the same working condition;

b. multiple test change trends of the same working condition;

c. Different working condition change conditions are tested at the same time;

d. c, a change trend in the process of multiple tests;

y＝ax+b (1)

Step 3.2.3, performing fault characteristic analysis:

y＝a₀t²+b₀t+c₀ (2)

Step 3.3, determining a threshold value by using the box diagram:

step 3.3.1, because the fault characteristics are that the values of the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are smaller and continuously reduced, the two values are respectively set as low thresholds, and the two thresholds of the oil inlet pressure of the supercharger and the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine under the conditions of 0.6, 0.8, 0.9 power factor, 20%, 50%, 75% and 90% load are respectively calculated by utilizing a box graph, wherein the threshold with larger value is defined as a pre-alarm value, and the threshold with lower value is defined as an alarm value;

step 4.2, a method for determining a threshold value is improved, the data of a fault unit are screened according to different active powers by using a programming language, a part of data with more operation times is screened, the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine under the same active power and different times of tests are respectively subjected to average treatment, two threshold values of the oil inlet pressure of the supercharger, the ratio of the oil inlet pressure of the supercharger to the oil pressure of the diesel engine are respectively obtained by using a box graph, wherein the threshold value with larger value is defined as a pre-alarm value, the threshold value with smaller value is defined as an alarm value, and the threshold value judgment of output quantities corresponding to different active powers under the same working condition can be realized without acquiring a power factor by the method;

P＝a₂T+b₂P_w+c₂ (3)

A total of 6 threshold expressions are finally determined, in the form of:

y₁＝a₁xⁿ+b₁x^n-1+… (4)

y₂＝a₁xⁿ+b₁x^n-1+… (5)

y₃＝a₂xⁿ+b₂x^n-1+… (6)

y₄＝a₂xⁿ+b₂x^n-1+… (7)

y₅＝a₁T+b₁P_W+c₁ (8)

y₆＝a₂T+b₂P_W+c₂ (9)

step 5.3, determining the degree of abnormality:

When W ₁ is less than W, the supercharger is lubricated normally;

Step 6, on-line prediction is realized;

Step 6.1, inputting the supercharger lubricating oil inlet temperature P, the supercharger lubricating oil inlet temperature T, the active power P _w and the diesel engine lubricating oil pressure P _D in real time through a data bus, preprocessing by utilizing a multi-threshold expression, outputting W, W ₁ and W ₂ under the current working condition, and entering a model to judge abnormal conditions;

step 6.2, outputting a prediction result: