A method for estimating thermal parameters of marine diesel engines for operating mode switching

Abstract

A method for estimating thermal parameters of a marine diesel engine for operating condition switching belongs to the field of thermal parameter identification of marine diesel engines, including: under different operating conditions, collecting thermal parameters such as speed, exhaust temperature, power, etc. during the operation of a diesel engine bench test bench and performing preprocessing; analyzing the relationship curves between the three variables of speed, exhaust temperature, and power and the load operating condition by polynomial fitting, and giving a fitting curve effect diagram, and evaluating the effect of the fitting by the mean absolute percentage error (MAPE); then analyzing the relationship between the three variables of speed, exhaust temperature, and power and the propulsion operating condition by polynomial fitting, and giving a fitting curve diagram, and also evaluating the effect of the fitting by the MAPE index, thereby realizing accurate estimation of thermal parameters of marine diesel engines under various operating condition changes. A method for estimating thermal parameters of a marine diesel engine for operating condition switching proposed by the present invention can improve the accuracy of parameter estimation, and is more in line with the actual working conditions of marine diesel engines. In addition, it can also be effectively applied to scenarios where the actual operating conditions are frequently switched. At the same time, the use of polynomial fitting can reduce the final model parameters, reduce the complexity of the model, and improve the calculation efficiency. Finally, the thermal parameter estimation method under common working conditions in this invention can be extended to the estimation of thermal parameters under any working conditions, and the method has strong scalability and practicality.

Description

Technical Field

[0001] The invention belongs to the field of thermal parameter identification of ship diesel engines and relates to a thermal parameter estimation method of ship diesel engines oriented to working condition switching. Background Art

[0002] Among various power equipment, diesel engines are widely used in the shipping industry due to their high thermal efficiency, good economy, strong adaptability and other advantages. They are the power source and heart of ships at sea. In recent years, ship diesel engines have gradually developed into the most commonly used ship power unit in the world. Among the existing large and medium-sized civilian ships, more than 90% use diesel engines as the main propulsion device, and only in some individual special fields do they use internal combustion engines, steam engines, etc. Due to the complex working process of diesel engines, there are many influencing factors, and a large number of thermal parameters are generated. In addition, these parameters also have the characteristics of many types of parameters, wide distribution range, and large amount of data. On the other hand, these large and widely distributed thermal parameters often carry a lot of key information and can reflect the operating status of diesel engines to a large extent. In the actual working scenario of diesel engines, due to the complex internal and external operating environment, the operating status of diesel engines is difficult to judge and estimate, which brings great hidden dangers to the safety and stability of diesel engines. By estimating the state of typical thermal parameters of a diesel engine, the operating state of the diesel engine can be inferred to a large extent. This method is simple, direct and reliable, and has important theoretical value and application prospects for ensuring the economical and high-quality operation of ship diesel engines. It is also an effective means to improve the intelligence level of ship power systems.

[0003] For multivariable complex systems such as marine diesel engines, most existing studies estimate or identify the parameters of a fixed operating condition. However, when a diesel engine is working, it is often accompanied by changes in the external environment and energy distribution requirements, which often requires the conversion of operating conditions according to different external conditions to achieve optimal efficiency. When the operating condition of the diesel engine is switched, the various thermal parameters of the diesel engine also change accordingly. At this time, the model established for a single operating condition will not be able to accurately give the estimated results of the target parameters under the changing operating conditions, thereby affecting subsequent research (such as fault diagnosis and early warning). How to establish an estimation model for typical thermal parameters under changing operating conditions not only has important theoretical value but also meets the actual needs of intelligent operation and maintenance of diesel engines. This patent proposes an effective method for estimating thermal parameters under variable operating conditions, which effectively solves the problem of parameter estimation of marine diesel engines after switching operating conditions according to actual needs, and has certain theoretical reference value and practical application prospects. Summary of the invention

[0004] The present invention is dedicated to solving the problem of estimating the main thermal parameters such as speed, exhaust temperature, power, etc. of a marine diesel engine when the operating conditions change. An effective thermal parameter estimation method for variable operating conditions is proposed, thereby realizing accurate estimation of thermal parameters of a marine diesel engine after the operating conditions change.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for estimating thermal parameters of marine diesel engines under variable operating conditions. The method first collects working data under different operating conditions generated by a bench test bench of a MAN 6S35ME-B9 diesel engine. Then, the relationship curves between the three variables of speed, exhaust temperature, and power and the changing load conditions are analyzed by polynomial fitting, and a fitting curve effect diagram is given. The effect of the fitting is evaluated by the mean absolute percentage error (MAPE) index. Then, the relationship curves between the three variables of speed, exhaust temperature, and power and the changing propulsion conditions are analyzed by polynomial fitting, and a fitting curve effect diagram is given. The effect of the fitting is also evaluated by the MAPE index, thereby achieving accurate estimation of thermal parameters of marine diesel engines when multiple major operating conditions change. Specifically, the following steps are included:

[0007] Step S1: Under different working conditions, data on the speed, exhaust temperature, and power variables of the diesel engine bench test bench during operation are collected and preprocessed.

[0008] In order to ensure the effectiveness of parameter fitting achieved by the method proposed in the present invention, it is necessary to process the state parameters of the diesel engine under different working conditions. The different working conditions specifically refer to: in order to better fit the relationship between the working condition and the speed, exhaust temperature, and power, and to make the fitting result closer to the actual state, different data of propulsion and load conditions are collected and processed respectively. In addition, since the exhaust temperature of the bench test often has abnormal points during the collection process (that is, the reading is significantly higher than the normal level), the value of the last detection time can be directly used for such abnormal data. After collecting and processing the state parameters under different working conditions, the monitoring variables can be obtained.

[0009] Step S2: Under the changing load conditions, a polynomial fitting is used to establish a relationship curve model between it and typical thermal parameters, and the MAPE index is used to evaluate the fitting effect:

[0010] Since the invention uses the working data generated by the diesel engine bench test bench for testing, in order to ensure that other process variables remain stable during the sampling process, when collecting working data under variable load conditions, the engine is set to work at a constant speed to simulate the situation where the ship speed remains unchanged under the ship load condition. Therefore, even when the diesel engine working condition changes, the speed of the diesel engine remains constant.

[0011] When fitting the relationship curve between exhaust temperature and load conditions, the process data collection time of 25% load condition is from 11:30:02 to 11:34:52 on May 14, 2016, the process data collection time of 50% load condition is from 11:45:02 to 11:49:52 on May 14, 2016, the process data collection time of 75% load condition is from 12:04:52 to 12:09:42 on May 14, 2016, and the process data collection time of 90% load condition is from 12:14:52 to 12:19:42 on May 14, 2016. The sampling interval is 10s, that is, 30 sample points are taken under each condition. The final fitting result curve is shown as follows:

[0012] T=4.1935*10 -4 t 3 -0.0811t 2 +5.9992t+161.4392

[0013] Where T is the exhaust temperature (degC), and t is the load condition. After calculation, the MAPE of the result curve and the actual monitoring data is 0.52%. The MAPE indicator expression is as follows:

[0014]

[0015] in, represents the estimated value of the fitted curve, y i Represents the actual measured value. MAPE is a percentage value that indicates the average deviation of the estimated result from the actual result. The smaller the MAPE value, the higher the accuracy of the parameter estimation.

[0016] When fitting the relationship curve between power and load conditions, the present invention collects power data at the same time point as the above load condition exhaust temperature data (25% load condition, 50% load condition, 75% load condition, 90% load condition), with a sampling interval of 10s, that is, 30 sample points are taken under each working state. The final fitting result curve expression is shown as follows:

[0017] P=-0.0073t 3 +1.2661t 2 -29.8399t+1014.3385

[0018] Wherein, P is power (kW), t is load condition. After calculation, the MAPE of the result curve and the actual monitoring data is 3.87%.

[0019] Step S3: Under the changing propulsion conditions, a polynomial fitting is used to establish a relationship curve model between it and typical thermal parameters, and the MAPE index is used to evaluate the fitting effect:

[0020] When fitting the relationship curve between the speed and the propulsion condition, the process data collection time of the 25% propulsion condition is from 12:59:52 to 13:06:02 on May 14, 2016, the process data collection time of the 50% propulsion condition is from 12:44:52 to 12:51:02 on May 14, 2016, the process data collection time of the 75% propulsion condition is from 12:29:52 to 12:36:02 on May 14, 2016, and the process data collection time of the 90% propulsion condition is from 12:14:52 to 12:21:02 on May 14, 2016, and the sampling interval is 10s. It is worth noting that except for the number of sample points under the 25% propulsion condition, which is 20 (the subsequent 18 sample points are directly discarded when the values ​​are 0), 38 sample points are taken under other working conditions. The final fitting result curve is shown as follows:

[0021] n=-0.0067t 2 +1.6392t+47.2291

[0022] Where n is the speed (r / min), and t is the propulsion condition. After calculation, the MAPE between the obtained result curve and the actual monitoring data is 14.6%.

[0023] When fitting the relationship curve between exhaust temperature and propulsion conditions, the present invention collects exhaust temperature data at the same time point as the above propulsion speed data (25% propulsion condition, 50% propulsion condition, 75% propulsion condition, 90% propulsion condition), the sampling interval is 10s, and 38 sample points are taken under each working state. The final fitting result curve is shown as follows:

[0024] T=5.1058*10 -4 t 3 -0.124t 2 +10.6154t+25.7013

[0025] Wherein, T is the exhaust temperature (degC), t is the propulsion condition. After calculation, the MAPE of the result curve and the actual monitoring data is 7.69%.

[0026] When fitting the relationship curve between power and propulsion conditions, the present invention collects power data at the same time point as the above propulsion condition speed data (25% propulsion condition, 50% propulsion condition, 75% propulsion condition, 90% propulsion condition), and the sampling interval is also 10s. Except for the number of sample points under the 25% propulsion condition, which is 19 (the values ​​of the subsequent 19 sample points are directly discarded as 0), 38 sample points are taken under other working conditions. The final fitting result curve is shown as follows:

[0027] P=-0.0165t 3 +2.8921t 2 -112.8566t+1960.5187

[0028] Wherein, P is power (kW), t is propulsion condition. After calculation, the MAPE of the result curve and the actual monitoring data is 2287%.

[0029] The curve fitting results between the power, speed variables and the changing propulsion conditions in this patent are not very ideal. The author believes that the reason may be that there is a direct shutdown in the later stage of the propulsion condition variable collection process, which causes some variables to undergo step changes. However, it can still meet the actual application needs, and for the remaining situations, this patent has achieved good parameter estimation theoretical results.

[0030] In summary, it is easy to find that the exhaust temperature and power parameters (the speed remains unchanged under load conditions and is not considered) generally increase with the increase of the diesel engine load. In fact, the greater the load during the operation of the diesel engine, the greater the force and torque required by the engine when doing work, and the higher the pressure and temperature generated by the explosion of the fuel in the cylinder, which is reflected in the exhaust temperature and power thermal parameters. The overall curve shows an upward trend with the increase of load conditions. On the other hand, in the conclusion of this patent, the speed, exhaust temperature, and power also increase with the increase of propulsion conditions. In reality, the increase in propulsion conditions directly leads to an increase in the amount of fuel supplied to the diesel engine and a larger opening of the throttle, resulting in an increase in the pressure in the combustion chamber, which in turn increases the torque and speed of the engine, accompanied by higher exhaust temperature and power. This shows that the typical thermal parameter estimation of this patent under variable conditions is consistent with the actual situation and has high practical value.

[0031] When using polynomials to establish a fitting model for variable working conditions, this patent only collects data under 25%, 50%, 75%, and 90% working conditions. In fact, if conditions permit, collecting data covering more and more complete working conditions for research will further improve the reliability of the experimental results. On the other hand, in practical applications, the model can be used to obtain an estimate of the target parameters for any working condition, so this method has strong scalability.

[0032] Compared with the prior art, the beneficial effects of the present invention are: it can improve the accuracy of parameter estimation, and is more in line with the actual working conditions of ship diesel engines, and can be effectively applied in scenarios where actual working conditions frequently switch; the use of polynomial fitting can reduce the final model parameters, reduce the complexity of the model, and improve calculation efficiency; the single thermal parameter estimation method in the invention can be extended to the estimation of any parameter, and the method has strong scalability and practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flow chart of the method for estimating thermal parameters of diesel engines for operating condition switching;

[0034] Figure 2 It is a graph showing the variation of exhaust temperature with load conditions;

[0035] Figure 3 It is a graph showing the power changing with load conditions;

[0036] Figure 4 It is a graph showing the speed changing with propulsion conditions;

[0037] Figure 5 is a graph showing the exhaust temperature changing with propulsion conditions;

[0038] Figure 6 It is a graph showing the power changing with propulsion conditions. DETAILED DESCRIPTION

[0039] The present invention is further described below in conjunction with specific embodiments of fitting the speed, exhaust temperature, and power when switching between load and propulsion conditions, but the protection scope of the present invention is not limited to the following embodiments. The complete implementation process includes the following steps:

[0040] Step S1: Collect the speed, exhaust temperature, and power data of the ship diesel engine under different working conditions and perform preprocessing. Specifically, this patent takes the MAN 6S35ME-B9 diesel engine as an example. For the load condition: collect the exhaust temperature and power data when the load condition is 25% from 11:30:02 on May 14, 2016 to 11:34:52 on May 14, 2016 Beijing time, and collect the exhaust temperature and power data when the load condition is 50% from 11:45:02 on May 14, 2016 to 11:49:52 on May 14, 2016 Beijing time. The exhaust temperature and power data at the time of the load condition were collected from 12:04:52 on May 14, 2016 to 12:09:42 on May 14, 2016 when the load condition was 75%, and the exhaust temperature and power data at the time of the load condition was 90% from 12:14:52 on May 14, 2016 to 12:19:42 on May 14, 2016. The sampling interval was 10s, that is, under each load condition, 30 sample points were taken for each propulsion condition: the speed, exhaust temperature and power data were collected from 12:59:52 on May 14, 2016 to 13:06:02 on May 14, 2016 Beijing time when the propulsion condition was 25%, the speed, exhaust temperature and power data were collected from 12:44:52 on May 14, 2016 to 12:51:02 on May 14, 2016 when the propulsion condition was 50%, and the speed, exhaust temperature and power data were collected from 20 The speed, exhaust temperature and power data when the propulsion condition was 75% were collected from 12:29:52 on May 14, 2016 to 12:36:02 on May 14, 2016. The speed, exhaust temperature and power data when the propulsion condition was 90% were collected from 12:14:52 on May 14, 2016 to 12:21:02 on May 14, 2016. The sampling interval was 10s, that is, 38 sample points were taken under each propulsion condition. In addition, for the abnormal points in the exhaust temperature of the bench test during the collection process (that is, the readings were significantly higher than the normal level), the value of the previous detection time was directly used instead.

[0041] Step S2: Under the changing load conditions, a polynomial fitting method is used to establish a relationship curve model between it and typical thermal parameters, and the MAPE index is used to evaluate the fitting effect. Since the speed of the bench test bench is constant under the load conditions, the engine speed does not change under the load conditions;

[0042] When fitting the relationship curve of exhaust temperature changing with load conditions, the exhaust temperature data under the changing load conditions in step S1 (the exhaust temperature data under all load conditions are taken at 30 sample points) are used to fit the relationship curve of exhaust temperature changing with load conditions, and the relationship expression of exhaust temperature changing with load conditions is obtained:

[0043] T=4.1935*10 -4 t 3-0.0811t 2 +5.9992t+161.4392

[0044] Where, T is the exhaust temperature (degC), and t is the load condition. The calculated MAPE = 0.52%, which indicates that the fitting result has good accuracy.

[0045] When fitting the relationship curve between power and load conditions, the power data under the changing load conditions in step S1 (power data under all load conditions are taken from 30 sample points) are used to fit the relationship curve between power and load conditions, and the relationship expression between power and load conditions is obtained:

[0046] P=-0.0073t 3 +1.2661t 2 -29.8399t+1014.3385

[0047] Among them, P is power (kW), t is load condition. After calculation, the result curve and the actual monitoring data have a MAPE of 3.87%, and the fitting effect is good.

[0048] Step S3: Under the changing propulsion conditions, a polynomial fitting is used to establish a relationship curve model between it and typical thermal parameters, and the MAPE index is used to evaluate the fitting effect.

[0049] When fitting the relationship curve between the speed and the propulsion condition, the speed data under the propulsion condition in step S1 (wherein the speed data of the first 20 sample points under the 25% propulsion condition and the speed data of 38 sample points under the other propulsion conditions) are used to fit the relationship curve between the speed and the propulsion condition, and the relationship expression between the speed and the propulsion condition is obtained:

[0050] n=-0.0067t 2 +1.6392t+47.2291

[0051] Where n is the speed (r / min), and t is the propulsion condition. After calculation, the MAPE of the result curve and the actual monitoring data is 14.6%, and the fitting effect is poor.

[0052] When fitting the relationship curve of exhaust temperature changing with propulsion conditions, the exhaust temperature data under the propulsion conditions in step S1 (38 sample points are taken for exhaust temperature data under all propulsion conditions) are used to fit the relationship curve of exhaust temperature changing with propulsion conditions, and the relationship expression of exhaust temperature changing with propulsion conditions is obtained:

[0053] T=5.1058*10 -4 t 3 -0.124t2 +10.6154t+25.7013

[0054] Wherein, T is the exhaust temperature (degC), and t is the propulsion condition. After calculation, the MAPE of the result curve and the actual monitoring data is 7.69%, and the fitting effect is average.

[0055] When fitting the relationship curve of power changing with propulsion conditions, the power data under the propulsion conditions changed in step S1 (wherein, the power data of the first 19 sample points are taken under the 25% propulsion conditions, and the power data of 38 sample points are taken under the other propulsion conditions) are used to fit the relationship curve of power changing with propulsion conditions, and the relationship expression of power changing with propulsion conditions is obtained:

[0056] P=-0.0165t 3 +2.8921t 2 -112.8566t+1960.5187

[0057] Wherein, P is power (kW), t is propulsion condition. After calculation, the MAPE of the result curve and the actual monitoring data is 2287%, and the fitting effect is very poor.

[0058] The above-described embodiments merely express the implementation methods of the present invention, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for estimating thermal parameters of a marine diesel engine for operating condition switching, characterized in that: The method collects key thermal parameters for different common working conditions, and then analyzes the relationship curves between the three typical thermal parameters of speed, exhaust temperature and power and the changing load conditions through polynomial fitting, and gives the fitting curve effect diagram. Then, the relationship curves between the speed, exhaust temperature, power and the changing propulsion conditions are analyzed through polynomial fitting, and the fitting curve effect diagram is given, so as to realize the thermal parameter estimation under the working condition switching condition; specifically, the following steps are included: Step S1: Under different working conditions, data of speed, exhaust temperature, and power variables during the operation of the diesel engine bench test bench are collected and preprocessed; The state parameters of the diesel engine under different working conditions are collected and processed; the different working conditions specifically refer to: in order to better fit the relationship between the working condition and the speed, exhaust temperature, and power, and also to make the fitting result closer to the real state, the data of different propulsion and load conditions are collected and processed respectively; in addition, since the exhaust temperature of the bench test often has abnormal points during the collection process, the readings of the abnormal points are significantly higher than the normal level, and the abnormal data of such abnormal points are directly replaced by the value at the last detection time; Step S2: Under the changing load conditions, a relationship curve model between the load and typical thermal parameters is established by using polynomial fitting, and the effect of the fitting is evaluated by the mean absolute percentage error (MAPE); In order to ensure that the remaining thermal parameters remain stable during the acquisition process, the engine is kept running at a constant speed when collecting process data under load conditions to simulate the situation where the ship speed remains unchanged when the ship is working under load; The relationship curve between the load condition and the exhaust temperature is fitted according to the changing load condition, and the results are as follows: T=4.1935*10-4t 3 -0.0811t 2 +5.9992t+161.4392 Among them, T is the exhaust temperature, the unit is degC, and t is the load condition; after calculation, the MAPE between the result curve and the actual monitoring data is 0.52%; The relationship curve between load condition and power is fitted according to the changing load condition, and the results are as follows: P=-0.0073t 3 +1.2661t 2 -29.8399t+1014.3385 Among them, P is power, unit is kW, t is working condition; after calculation, the result curve and the actual monitoring data have MAPE = 3.87%; Step S3: Under the changing propulsion conditions, a relationship curve model between the polynomial fitting and typical thermal parameters is established, and the MAPE index is used to evaluate the fitting effect; The relationship curve between the changing propulsion conditions and the speed is fitted, and the results are as follows: <h2 style=";text-align:left;direction:ltr">n = -0.0067t<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> +1.6392t+47.2291 Among them, n is the speed, the unit is r / min, and t is the propulsion condition. After calculation, the MAPE of the result curve and the actual monitoring data is 14.6%; The relationship curve between the changing propulsion conditions and the exhaust temperature is fitted, and the results are as follows: T=5.1058*10 -4 t 3 -0.124t 2 +10.6154t+25.7013 Wherein, T is the exhaust temperature, the unit is degC, and t is the propulsion condition. After calculation, the MAPE of the result curve and the actual monitoring data is 7.69%. In addition, the polynomial fitting method can reduce the final model parameters, reduce the complexity of the model, and improve the calculation efficiency. The relationship curve between the changing propulsion conditions and power is fitted, and the results are as follows: P=-0.0165t 3 +2.8921t 2 -112.8566t+1960.5187 Among them, P is power, unit is kW, t is propulsion condition; after calculation, the result curve obtained is MAPE=2287% with the actual monitoring data; In the steps S2 and S3: The polynomial fitting is to calculate the linear combination of the high-order terms of the input variable x, where the highest cubic term of the input variable is used. The specific expression is as follows: in, represents the estimated value based on the input variable x and parameter w, w i Represents the target parameter; the resulting model can operate as efficiently as a linear model, while making the model applicable to all working conditions; the sum of squared errors is used as the loss function, and the expression is as follows: Among them, E(w) represents the result loss, N represents the number of test samples, and x n represents the nth test sample data, t n Represents the actual label of the nth test sample. When the loss function takes the minimum value, the optimal estimation model is obtained.

Images