JPH01210840A - Abnormality diagnostic expert system for diesel engine - Google Patents

Abstract

PURPOSE:To detect abnormality of an engine in an on-line real time in an early stage by constituting the title system so that pressure waveform data obtained by always measuring combustion pressure in a cylinder is collated with a standard pattern of a normal or abnormal waveform which is stored in advance. CONSTITUTION:Combustion pressure data of each cylinder is obtained by a pressure detecting part 3, and this data is brought to amplification switching 6 and inputted to a data processing part 7. Subsequently, in accordance with a power generation load state, a deviation value of this data is calculated, and by comparing the deviation value and a prescribed value, whether abnormality exists or not is discriminated. When the deviation value does not coincide with the prescribed value, an alarm signal is outputted, and pressure data, etc. of the cylinder in which abnormality is detected are outputted to an inferring part 8 from the processing part 7. Next, the inferring part 8 specifies which feature value whose standard pattern is to be checked, and compares whether a feature value of the detected pressure waveform, for instance, the maximum pressure goes into a prescribed range of the maximum pressure of the standard pattern which is stored in advance or not. As a result of comparison, when said feature value is in the prescribed range, weighting by a normal distribution is executed and an evaluation value is calculated.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an expert system for diagnosing abnormalities using combustion pressure waveform data in each cylinder of a diesel engine.

[Prior Art] Conventionally, there has been no particular system for automatically diagnosing abnormalities in diesel engines, and it has been common practice to measure and monitor the pressure, temperature, flow rate, etc. around the engine. An example of a device used for monitoring such an engine is the product name rED.
There is a device called EN ST-30. This is a remote type monitoring device that allows monitoring of marine engines from the bridge. This device is not a device for measuring combustion pressure in a cylinder, and therefore does not have a permanent cylinder pressure detection section. Furthermore, it is not an expert system for diagnosing diesel engine abnormalities, but rather a device that simply monitors the conditions around the engine.

Furthermore, there is a device with the trade name rcOMOs D4J. This is aimed at marine engines and is equipped with a pressure detection section to measure cylinder combustion pressure, but it is a device used in engine efficiency calculations, etc., and is not an abnormality diagnosis system based on cylinder combustion pressure data. Additionally, the cost of the equipment is high.

[Problems to be solved by the invention] By the way, diesel engines used for power generation on remote islands are required to have high reliability due to their public nature, and it is important to detect abnormalities in such diesel engines early. It is necessary to prevent failures before they occur. however,
There is no abnormality diagnosis system for diesel engines that meets these requirements. The reality is that a device slightly like the one described above is used in marine diesel engines, but in such a device, each cylinder (1 There is a problem in that it is physically almost impossible to constantly monitor the combustion state of the engine (average of 18 cylinders).

In view of the above-mentioned problems with the conventional type, the present invention provides a diesel engine abnormality diagnosis expert system that is capable of always online detecting the internal cylinder pressure of a diesel engine, diagnosing its abnormality, and outputting the information to the outside. The purpose is to provide.

[Means for Solving the Problems] In order to achieve the above object, the diesel engine abnormality diagnosis expert system according to the present invention includes means for detecting the internal cylinder pressure provided for each cylinder of the diesel engine; means for acquiring pressure data of each cylinder over time from the detection means; means for extracting predetermined characteristic values based on the pressure data; and characteristic values of the plurality of standard patterns stored in advance in conjunction with the characteristic values. means for calculating an evaluation value for evaluating the mutual similarity between the pattern of cylinder pressure change and the standard pattern, and means for inferring and outputting the content of the abnormality of each cylinder based on the evaluation value. It is characterized by having the following.

[Operation] With the above configuration, abnormality diagnosis of the diesel engine is performed as follows. That is, first, the combustion pressure in each cylinder of a diesel engine is detected online at all times, data indicating the pressure change is acquired over time, a predetermined characteristic value is extracted, and the characteristic value and the pre-stored An evaluation value for evaluating the mutual similarity between the pressure change pattern and the standard pattern is calculated by comparing the characteristic values of multiple standard patterns, and the content of the abnormality of each cylinder is determined based on the evaluation value. Infer and output.

[Example] Hereinafter, the present invention will be explained in detail using the drawings.

FIG. 1 is a block diagram showing the configuration of a diesel engine abnormality diagnosis expert system according to an embodiment of the present invention. In the figure, 1 is a diesel engine, which includes a plurality of cylinder blocks 2. As shown in FIG. Each cylinder block 2 has an in-cylinder combustion pressure detection section 3.
is provided, which detects the combustion pressure within the cylinders of each cylinder block 2 and extracts it as an analog electrical signal.

4 is a shaft of the diesel engine 1 that transmits power to the generator 12. This shaft 4 is equipped with a crank angle detection section 5, which detects the crank angle corresponding to each cylinder based on the rotational position of the shaft 4 and can output it as a digital electrical signal.

Reference numeral 6 denotes a signal amplification/switching section which amplifies the minute combustion pressure electric signal of each cylinder block 2 sent from each pressure detection section 3 and outputs it to the data processing section 7 while switching between a plurality of cylinders. The data processing section 7 processes the analog signal from the pressure detection section 3 sent via the signal amplification switching section 6 and the digital signal from the crank angle detection section 5h1, etc., and processes the maximum pressure in the cylinder for each cylinder block 2. , average effective pressure, average output, rotation speed, etc. are calculated and output.

The pressure detection section 3, crank angle detection section 5, signal amplification switching section 6, and data processing section 7 described above are collectively referred to as a cylinder pressure monitoring section. In this embodiment, the cylinder pressure monitoring unit is a cylinder pressure monitoring device manufactured by ASEA (headquartered in Sweden) (the distributor in Japan is Gabrius Marine Co., Ltd.), and the interface of the data processing unit 7 is modified. using things. Note that conventional detection units for measuring the pressure inside the cylinder have had problems in terms of durability, accuracy, and the like. However, the cylinder pressure monitoring device used in this example has sufficient durability and long life to continuously measure the cylinder pressure while the diesel engine is operating. In addition, as will be described later, in order to quickly infer and discover abnormalities in the diesel engine body based on the data measured by the pressure detection unit 3, the measurement data that serves as the basis for judgment must have sufficient accuracy. It is a device that is satisfactory in all respects.

8 is an inference section, 9 is a CRT, 10 is a printer, and 11 is a hard disk. In this embodiment, an existing personal computer is used as the inference section 8.

Next, the data sent from the data processing section 7 to the inference section 8 will be explained in detail. First, cylinder internal pressure data is sent out. This is cylinder pressure data at each time point during one revolution of the crank (engine compression stroke to expansion stroke) divided into 800 equal intervals. Furthermore, the data processing unit 7 calculates the data of the feature values as shown below.
It is sent to the inference section 8.

■ Maximum pressure ■ Average effective pressure ■ Pressure at points T, D, and ■ Expansion pressure ■ Number of revolutions ■ Maximum pressure period and angles of T, D, and G ■ Output horsepower and total output ■ Peak value of waveform (maximum pressure Diagram) Basic data for waveform classification will be explained with reference to the waveform diagram in FIG. By sequentially plotting the above-mentioned 800 cylinder internal pressure data over time, a waveform diagram of the cylinder internal pressure with respect to the rank angle as shown in the figure can be obtained. In the same figure, 2
1 is a compression curve showing the change in cylinder pressure during non-combustion. Data during non-combustion is measured and stored in advance for each cylinder, such as when starting the diesel engine. 22
is a combustion curve showing the change in cylinder internal pressure after combustion starts. 23 is the top dead center of the crank (T, -D,
(abbreviated as C). 24 indicates the point at which the pressure of the combustion curve 22 reaches its peak.

The maximum pressure of the characteristic value ■ mentioned above is the maximum pressure of the combustion curve 22. The average effective pressure of the characteristic value (■) is the average value of the area of the combustion curve 22 minus the area of the non-combustion curve. The ignition timing of the characteristic value ■ is the ignition timing shown in ■ in the figure. The characteristic value ■ is the maximum pressure period of ■ in the figure and T,
These are the crank angles of D and C.

The inference unit 8 inputs these data and compares them with each feature value of a plurality of standard patterns of combustion waveforms. Third
Figures (a) to (i) are waveform diagrams of standard patterns.

The figure also shows the compression curve during non-combustion for comparison. The figure (a) shows the normal waveform (pattern 0),
Figure (b) shows the waveform (pattern 1) when a fuel valve abnormality occurs due to valve adhesion or wear, or poor spring adjustment or deterioration, resulting in poor fuel spray conditions. Figure (C) shows a waveform when a leak occurs. (Pattern 2) shows the waveform when an abnormality occurs in the fuel injection pump system due to a problem with the injection pump or fuel, and (d) shows the waveform when a pressure abnormality with two peaks in the pressure waveform occurs due to a problem with the injection pump or fuel. The waveform (pattern 3), (e) is the waveform (pattern 4) when an abnormality occurs in which the fuel injection timing is too early, and (f) in the same figure shows the waveform (pattern 4) when an abnormality in which the fuel injection timing is too late occurs. Pattern 5), (g) is the waveform when an abnormality occurs where the cylinder compression pressure decreases (pattern 6), and (h) is the waveform when an abnormality occurs where irregular irregularities appear on the combustion curve. (Pattern 7), Figure (i) shows the waveform when knocking occurs (Pattern 8).
) is shown.

The above-mentioned characteristic values (1) to (2) for these standard patterns are calculated and stored in advance. Then, the characteristic values of the actually measured combustion pattern are compared with the characteristic values of the standard pattern, and an evaluation value indicating which standard pattern is similar to and to what degree is calculated.

Next, with reference to the flowchart of FIG. 4, the procedure for diagnosing an abnormality by the system of the present embodiment shown in FIG. 1 will be explained.

In this system, first, in step S1, the pressure detection unit 3 generates combustion pressure data (analog signal) for each cylinder.
get. This data is input to the data processing section 7 via the signal amplification switching section 6, and is subjected to analog/digital conversion in step S2. Next, in step S3, the deviation value of this pressure data is calculated according to the power generation load state, and step S3
In step 4, the deviation value is compared with a specified value to determine whether there is an abnormality. If there is no abnormality, the data processing unit 7
issues a command to detect the combustion pressure of the next cylinder, and then returns to step S1 to repeat this operation sequentially at all times on-line.

Next, if the deviation value deviates from the specified value in step S4, the process proceeds to step S6, and an alarm signal is output as an abnormality. Furthermore, in step S7, pressure data and the like of the cylinder in which the abnormality has been detected is outputted from the data processing section 7 to the inference section 8. The above processing of steps 51 to S7 is performed by the cylinder pressure monitoring section.

Next, in step S8, the inference unit 8 specifies which feature value of which standard pattern is to be checked. That is, for example, checking the maximum pressure of the characteristic value (3) of the normal pattern shown in FIG. 3(a) is specified. Next, in step S9, the characteristic value of the detected pressure waveform data, such as the maximum pressure, is stored in advance in FIG. 3 (a).
) is compared to see if it falls within a predetermined range of the maximum pressure of the standard pattern. Note that each feature value of each standard pattern is specified within a certain predetermined range.

As a result of the comparison in step S9, if the feature value is within the range of the feature value of the standard pattern, in step SIO weighting is performed using a normal distribution and an evaluation value is calculated.

This normal distribution weighting is applied to the standard pattern when the range of characteristic values of the standard pattern is, for example, from the minimum value a to the maximum value, and the characteristic value of the measured pressure waveform pattern is C (a<cab). The weight value corresponding to C is obtained assuming that the proportion of similarity to the pattern is normally distributed between a and b. This weight value is multiplied by a constant and added to the evaluation value (initial value 0). The constant to be multiplied indicates the weighting between each feature value. For example, when evaluating the similarity with a standard pattern, if the feature value ■ shows a clearer degree of similarity than the feature value ■, the constant of the feature value ■ is increased and the weighting is set to dog. .

On the other hand, as a result of the comparison in step S9, if the feature value is outside the range of the feature values of the standard pattern, the similarity evaluation based on the feature value is 0, so the evaluation value is left unchanged and the process proceeds to step S11. . Next, in step Sit, it is checked whether comparison has been completed for all feature values of the standard pattern. If all the comparisons have not been completed yet, the process returns to step S8 and checks the next feature value of the standard pattern.The comparison is performed in step S9, and if the result is within the range, weighting is performed in step SIO. If it is outside the range, the evaluation value is left as is, and it is checked again in step Sll whether or not all features have been calculated. Through such loop processing, comparisons are completed for all feature values of one standard pattern, and an evaluation value indicating the degree of similarity with the standard pattern is obtained. Note that the ratio of the evaluation value calculated as above to the value previously calculated and stored by taking the center point of the range of each feature value of the standard pattern (hereinafter referred to as
(referred to as the evaluation value ratio). This evaluation value ratio is 0 or more and 1
When the value is 1, it can be seen that the degree of similarity to the standard pattern is extremely high.

Next, when calculation has been completed for all the features of the standard pattern in step Sit, it is determined in step S12 whether calculation has been completed for all the standard patterns. if,
If there is a standard pattern that has not been calculated yet, go to step S.
Returning to step 8, the same calculation as above is performed for the next standard pattern to calculate the evaluation value and evaluation value ratio. When calculations are completed for all standard patterns, the top three evaluation value ratios are output in step S13, and further, in step S14, the abnormal location where an abnormality is inferred, the possible causes of the abnormality, and the main measures to be taken are output to the CRT 9. and exit.

FIG. 5 shows an example of a screen display output to the CRT 9. In the figure, the cylinder targeted for abnormality diagnosis is identified by displaying it as "No. 2 machine, No. 5 cylinder." Furthermore, the date, time, etc. are also displayed. The pressure waveform is displayed by changing the line color of a compression curve and a combustion curve during non-combustion, which have been obtained in advance.

The three evaluation value ratios are displayed in order from the one with the largest value.

In this example, the degree of similarity to pattern 0 (indicating normality) is the highest at 0.95, followed by pattern 6 and pattern 4.
The degree of similarity between the two figures is 0.08 and 0.03, respectively. Further, these data can also be output to the printer 10 and the hard disk 11. It is for convenience that the data to be outputted is set to the top three, and can be changed as appropriate.

In the above embodiment, as shown in FIG. 1, an example of the configuration is shown in which abnormality diagnosis is performed for a plurality of cylinders of one diesel engine, but the same configuration can be applied even when there are a plurality of diesel engines. In that case, each cylinder may be provided with a pressure detection section 3, and each diesel engine 1 may be provided with a crank angle detection section 5. Further, although patterns 0 to 8 are set as standard patterns, the present invention is not limited to this, and any pattern can be used as long as it can be discriminated by characteristic values.

Furthermore, as for characteristic values, we focused on characteristic values related to cylinder pressure such as those mentioned above, but we also focused on characteristic values related to cylinder pressure, such as engine speed, exhaust gas temperature, vibration, sound, cooling water inlet/outlet temperature, and lubricating oil. If we focus on the viscosity, temperature, etc., and detect these using appropriate sensors and process them in the data processing section and reasoning section, more detailed abnormality diagnosis can be performed.

Furthermore, according to the above embodiment, the evaluation data as described above can be automatically acquired for each cylinder of each diesel engine, so if these data are stored for a long period of time, the data will not change over time. Changes can be monitored. In other words, for example, it is easy to obtain data that shows that a certain cylinder is gradually approaching a certain abnormal pattern, so by inspecting such a diesel engine in advance, you can prevent it from shutting down due to a sudden failure. It is possible to avoid such situations.

[Effects of the Invention] As explained above, according to the present invention, the pressure waveform data obtained by constantly measuring the combustion pressure in the cylinder is compared with a standard pattern of normal or abnormal waveforms stored in advance. Since abnormalities in all cylinders are inferentially diagnosed, the following effects are achieved.

(1) Diesel engine abnormalities can be detected early online in real time.

(2) Since the inspection timing of the diesel engine can be predicted, the number of inspections can be reduced and effective periodic inspections can be performed.

(3) Based on the results of diagnosis, the inspection scope and contents during inspection can be specified in detail, preventing unnecessary parts replacement and reducing parts inventory.

(4) Since the combustion state of each cylinder can be checked at all times, any decrease in engine efficiency can be detected at an early stage.

(5) After engine repair, it is possible to check whether combustion conditions and timing adjustments have been made properly.

(6) Conventionally, one or two people walked and inspected the engine surroundings every hour or so, but according to the present invention, this labor can be reduced.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a diesel engine abnormality diagnosis expert system according to an embodiment of the present invention. FIG. 2 is a waveform diagram for explaining basic data for waveform classification. ) to (i) are standard pattern waveform diagrams, 4th
The figure is a flow sheet for explaining the procedure of abnormality diagnosis by the system of this embodiment, and FIG.
It is an explanatory diagram showing a mode displayed on an R7 screen. 1: diesel engine, 2: cylinder block, 3: pressure detection section, 5: crank angle detection section, 6: signal amplification switching section, 7: data processing section, 8: reasoning section. (0) Normal (pattern 0) (C) Abnormality of LW piece/response 1 inch pump 7L (pattern 2)
) Fig. 3 (d) Twin Peak Plate 7v-() Dimension 3) <
e>) Prisoner's hat! Abnormal period (pattern 4)
) (f) Instigation and banquet 1 sun-? IK (J) abnormality (pattern 5, Figure 3 (9) cylinder, compression piston (pattern 6) (h
) unwritten tll'l? 7 majors, 1 exemption () Makoon 7) (+)
Notsking (pattern 8) Figure 3

Claims (1)

[Scope of Claims] Means for detecting the pressure inside the cylinder provided for each cylinder of a diesel engine; means for acquiring pressure data of each cylinder over time from the detecting means; a means for extracting characteristic values of the cylinder; and comparing the characteristic values with characteristic values of a plurality of pre-stored standard patterns to evaluate the mutual similarity between the pattern of cylinder pressure changes and the standard patterns. An expert system for diagnosing an abnormality in a diesel engine, comprising means for calculating an evaluation value, and means for inferring and outputting the content of an abnormality in each cylinder based on the evaluation value.