CN115753455A - Electric drive system multidimensional load prediction method based on model and data hybrid driving - Google Patents
Electric drive system multidimensional load prediction method based on model and data hybrid driving Download PDFInfo
- Publication number
- CN115753455A CN115753455A CN202211369253.7A CN202211369253A CN115753455A CN 115753455 A CN115753455 A CN 115753455A CN 202211369253 A CN202211369253 A CN 202211369253A CN 115753455 A CN115753455 A CN 115753455A
- Authority
- CN
- China
- Prior art keywords
- model
- drive system
- electric drive
- load
- failure
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000013528 artificial neural network Methods 0.000 claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 24
- 238000011156 evaluation Methods 0.000 claims abstract description 12
- 230000007246 mechanism Effects 0.000 claims abstract description 10
- 230000004927 fusion Effects 0.000 claims abstract description 9
- 230000035882 stress Effects 0.000 claims description 21
- 238000009826 distribution Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 10
- 230000032683 aging Effects 0.000 claims description 8
- 238000004088 simulation Methods 0.000 claims description 7
- 238000005050 thermomechanical fatigue Methods 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 5
- 230000002068 genetic effect Effects 0.000 claims description 5
- 238000004804 winding Methods 0.000 claims description 5
- 230000015556 catabolic process Effects 0.000 claims description 4
- 238000006731 degradation reaction Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 229910000831 Steel Inorganic materials 0.000 claims description 3
- 238000012821 model calculation Methods 0.000 claims description 3
- 239000010959 steel Substances 0.000 claims description 3
- 238000012795 verification Methods 0.000 abstract description 4
- 238000012827 research and development Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000012544 monitoring process Methods 0.000 description 6
- 239000000463 material Substances 0.000 description 5
- 238000012360 testing method Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000005452 bending Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical group [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000003878 thermal aging Methods 0.000 description 1
- 230000008646 thermal stress Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Abstract
The invention discloses a multi-dimensional load prediction method of an electric drive system based on model and data hybrid driving, which comprises the following steps: acquiring actual user operation data, and analyzing failure mechanisms and failure associated loads of different components of the electric drive system based on the actual user operation data; extracting a multidimensional failure associated load by adopting a model driving method; constructing a multidimensional load rapid prediction model of the electric drive system based on neural network algorithm fusion by adopting a model and data hybrid drive method; and (3) constructing a load prediction precision evaluation index, and verifying the effectiveness of the multi-dimensional load rapid prediction model of the electric drive system. The method can realize the prediction of the failure associated load of the electric drive system on different time scales such as key stress, power loss, time delay temperature and the like, greatly reduce the cost for acquiring the failure associated load, provide support for real-time evaluation of damage of each part of the electric drive system and dynamic prediction of service life, and simultaneously provide a reliable data source for product research and development and verification links.
Description
Technical Field
The invention relates to the technical field of health monitoring of electric drive systems, in particular to a multidimensional load prediction method of an electric drive system based on model and data hybrid driving.
Background
The electric drive system is used as a highly integrated electromechanical liquid integrated product, bears complex dynamic alternating multi-physical-field load in actual work, and different component failure mechanisms and failure associated loads have differences. The method has the advantages that the multi-dimensional load signals related to the failure of each part are accurately and rapidly acquired, and the method has important significance for monitoring the damage state of the electric drive system and predicting the residual life. At present, a method for collecting loads is a direct method for installing multiple sensors, but the problems of high cost, limited installation position, high data storage cost and the like are faced, and associated load data under multiple failure modes are difficult to obtain comprehensively; the method can accurately acquire the related load based on a physical model simulation or analysis model, but the time consumption is long, and the prediction precision depends on model parameters; considering the data-driven approach alone can establish a non-linear mapping relationship between input and output signals, but lacks the interpretability of the correlation mechanism. Therefore, the invention provides a multi-dimensional load prediction method of an electric drive system based on model and data hybrid driving, so as to realize rapid prediction of failure associated loads of the electric drive system under different time scales.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a multidimensional load prediction method of an electric drive system based on model and data hybrid driving, which can realize the prediction of failure associated loads of different time scales such as key stress, power loss, time delay temperature and the like of the electric drive system, greatly reduce the cost for acquiring the failure associated loads, provide support for real-time evaluation of damage and dynamic prediction of service life of each component of the electric drive system, and provide a reliable data source for product research and development and verification links.
In order to achieve the technical purpose, the invention provides a multidimensional load prediction method of an electric drive system based on model and data hybrid driving, which comprises the following steps:
acquiring actual user operation data, and analyzing failure mechanisms and failure associated loads of different components of the electric drive system based on the actual user operation data;
extracting a multidimensional failure associated load time domain signal by adopting a model driving method;
constructing a multi-dimensional failure associated load rapid prediction model of the electric drive system based on neural network algorithm fusion by adopting a model and data hybrid drive method;
and constructing a load prediction precision evaluation index, and verifying the effectiveness of the quick prediction model of the multidimensional failure associated load of the electric drive system.
Optionally, the actual user operation data includes, but is not limited to, vehicle speed, motor torque, motor current, and motor voltage acquired by the electric drive system during actual operation;
the different components of the electric drive system include, but are not limited to, a motor shaft, gears, bearings, stator windings, rotor magnetic bridges, rotor magnetic steel, power devices IGBT.
Optionally, the failure mechanism is a failure mode and a failure physical model of the different components of the electric drive system under multi-dimensional excitation load;
the failure modes include, but are not limited to, fatigue failure, wear failure, aging failure, thermomechanical fatigue failure;
the physical model of failure includes, but is not limited to, the Basquin model under the fatigue failure, the Archard wear model under the wear failure, the Arrhenius degradation model under the aging failure, the Coffin-Manson model under the thermomechanical fatigue failure;
the failure-related load is the load associated with causing the component to fail, including but not limited to motor speed, motor torque, motor current, motor voltage, contact stress, power loss, temperature.
Optionally, the model-driven method includes, but is not limited to, a load acquisition method based on physical model simulation and a load acquisition method based on analytical model calculation.
Optionally, the model and data hybrid driving method includes:
and on the basis of the multi-dimensional failure associated load acquired by the model driving method, constructing a multi-dimensional load rapid prediction model of the electric drive system based on algorithm fusion of a GA (genetic algorithm) optimized BP (Back propagation) neural network and an LSTM (least Square TM) neural network by adopting a data driving method, and optimizing the structure and parameters of the multi-dimensional load rapid prediction model of the electric drive system.
Optionally, the algorithm is fused as:
and on the basis of a GA-BP algorithm, taking the predicted power loss as a main input signal of a temperature prediction model, and combining an original input layer signal in the GA-BP algorithm and the predicted power loss to construct a neural network intermediate layer, wherein the neural network intermediate layer is used as an input layer of an LSTM neural network, so that the multi-dimensional load rapid prediction model of the electric drive system is obtained.
Optionally, the optimization includes, but is not limited to, implicit layer number optimization of the BP neural network, implicit layer node number optimization, initial weight and threshold optimization, and initial parameter optimization of the LSTM neural network.
Optionally, the load prediction accuracy evaluation index includes, but is not limited to, a correlation coefficient, a root mean square error, a time-domain frequency distribution, and a damage recurrence ratio;
the correlation coefficient is larger than 0.9, the difference of the time domain frequency distribution is smaller than 10%, and the damage recurrence ratio is between 0.8 and 1.2.
Optionally, the root mean square error is calculated by:
the calculation formula of the correlation coefficient is as follows:
wherein, RMSE is root mean square error; r 2 Is a correlation coefficient; y is i 、
Respectively a target value and a predicted value;
is the average of the target values; n is the number of samples;
the calculation formula of the damage recurrence ratio is as follows:
in the formula, delta i As the damage reproduction ratio of the component i,
for the damage value of component i under the predicted load,
is the damage value of component i at the desired target load.
The invention has the following technical effects:
the method comprises the steps of analyzing failure mechanisms and failure associated loads of different parts of the electric drive system based on actual user operation data, extracting multi-dimensional failure associated load time domain signals through a model driving method, verifying the effectiveness of the model driving method by combining actual test data, constructing a multi-dimensional load rapid prediction model of the electric drive system based on the fusion of a neural network algorithm based on a model and data hybrid driving method, constructing a reasonable load prediction precision evaluation index, and analyzing the effectiveness of the established prediction model from multiple angles. According to the technical method, the failure associated load prediction of different time scales such as key stress, power loss, time delay temperature and the like of the electric drive system can be realized, the cost for acquiring the failure associated load is greatly reduced, support is provided for real-time evaluation of damage and dynamic prediction of service life of each part of the electric drive system, and meanwhile, a reliable data source can be provided for product research and development and verification links.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of a multidimensional load prediction method of an electric drive system based on model and data hybrid driving according to an embodiment of the invention;
FIG. 2 is a schematic diagram of a multi-dimensional load prediction model based on GA-BP-LSTM algorithm fusion according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a prediction error of a BP network model under different numbers of hidden layer nodes according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a BP network model prediction error under different crossover and variation probabilities according to an embodiment of the present invention;
FIG. 5 is a schematic diagram illustrating a time domain comparison of stress and critical loss prediction results according to an embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating time domain comparison of different temperature prediction results according to an embodiment of the present invention;
FIG. 7 is a schematic diagram illustrating a comparison between predicted values of different loads and expected frequency distributions according to an embodiment of the present invention;
FIG. 8 is a graph illustrating predicted damage versus expected damage for various components of an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in FIG. 1, the invention discloses a multidimensional load prediction method of an electric drive system based on model and data hybrid driving, which comprises the following steps:
s1, acquiring actual user operation data, and analyzing failure mechanisms and failure associated loads of different components of the electric drive system based on the actual user operation data. Under the multi-source excitation load, multiple failure modes such as fatigue, abrasion, aging, thermal mechanical fatigue and the like exist in all parts of the electric drive system, the load transfer rule of all the parts of the electric drive system is researched, and support is provided for load prediction and damage monitoring.
Specifically, in this embodiment, the actual user operation data in S1 includes, but is not limited to, a vehicle speed, a motor torque, a motor current, and a motor voltage, which are obtained in an actual operation process of the electric drive system; the different parts of the electric drive system include but are not limited to motor shaft, gears, bearings, stator windings, rotor magnetic bridge, rotor magnetic steel, power devices IGBT, etc. The failure mechanism is a failure mode and a failure physical model of different components of the electric drive system under multi-dimensional excitation load; failure modes include, but are not limited to, fatigue failure, wear failure, aging failure, thermomechanical fatigue failure, including in particular: failure modes of different parts under multi-source excitation load are different, and torsional fatigue failure is easy to occur due to shear stress generated by a motor shaft under the condition of bearing alternating torque; the gear parts are easy to cause contact fatigue failure and bending fatigue failure under the alternating load of rotating speed and torque; the bearing part is influenced by radial force, axial force and centrifugal force under the alternating load of rotating speed and torque, so that fatigue and wear failure are easily caused; the motor rotor is subjected to mechanical stress and thermal stress under the working conditions of high dynamic alternating rotating speed and torque, so that fatigue failure is easy to generate; under the operating condition, the temperature inside the stator and the rotor changes due to loss change, and the continuous temperature load can cause thermal aging failure on the stator and rotor iron cores, windings and the like; the power device IGBT module is easy to generate thermal mechanical fatigue and aging failure under the working condition of temperature alternation.
The physical model of failure specifically includes: different failure physical models can be adopted to carry out life assessment under different failure modes, for example, a Basquin life model can be adopted to carry out life assessment under fatigue failure; an Archard wear model can be adopted under the wear failure; an Arrhenius degradation model can be adopted under aging failure; a Coffin-Manson model and the like can be adopted under the thermo-mechanical fatigue failure. The damage calculation model includes but is not limited to the linear accumulated damage criterion, the bilinear accumulated damage criterion, the nonlinear accumulated damage criterion, etc.;
the calculation formula of the Basquin service life model is as follows:
ΔS m N＝C
in the formula, Δ S is a stress amplitude, N is a fatigue life under the action of the stress amplitude S, and m and C are constants related to a material.
The calculation formula of the Archard wear model is as follows:
wherein W is a wear volume, s is a frictional sliding distance, P is an applied load, and P is m Is the flow pressure of the material and K is the wear coefficient between the materials.
The calculation formula of the Arrhenius degradation model is as follows:
L＝A·e -E/kT
where L is lifetime, A is a constant, frequency factor, E is activation energy, material dependent, in eV, k is Boltzmann constant, and T is temperature stress in Kelvin. The calculation formula of the coffee-Manson model is as follows:
N f ＝Af -α ΔT -β G(T max )
in the formula, N f For part life, G (T) max ) The energy of activation of Arrhenius at the highest temperature stress, Δ T is the temperature difference between the highest temperature and the lowest temperature in Kelvin, f is the cycle frequency in Hertz, and A, α, and β are constants.
S2, extracting the multidimensional failure associated load by adopting a model driving method;
specifically, in this embodiment S2, based on load conditions, such as vehicle speed, motor rotation speed, motor torque, motor current, and motor voltage, that are easily obtained in the actual operation process of the electric drive system, a model driving method is used to establish a physical simulation model or an analytic model for obtaining failure loads of different components, and multidimensional failure-related loads, such as critical stress, power loss, and delay temperature, in the electric drive system are extracted by combining actual materials, structures, and performance parameters of the components. And verifying the effectiveness of the established physical simulation model or analytical model through the corresponding bench test monitoring response signal.
Model-driven methods include, but are not limited to, load acquisition methods based on physical model simulation and load acquisition methods based on analytical model calculations. Load data such as motor shaft shear stress, gear contact stress, gear bending stress, bearing radial force and axial force, coupling stress of a rotor magnetic bridge under multiple physical fields, motor stator and rotor loss, motor stator and rotor temperature, power device IGBT loss and temperature and the like in an electric drive system can be obtained through a model driving method based on actual operation conditions. The effectiveness of the model driving method is verified through real test data, the method is long in time consumption, and accurate failure associated load data under different user working conditions can be obtained, so that a training set of actual operation user working condition data and failure associated load data is established and used in a data-driven load rapid prediction model.
In the verification of the real test data, the fact that loads such as loss, rotor stress, temperature and the like in multiple physical fields are difficult to acquire through sensors is considered, but the temperature distribution trends of adjacent parts are basically consistent, whether the temperature distribution rule of the rotor is reasonable or not can be verified by monitoring the temperature of the stator through monitoring loads of adjacent parts through arranging thermistors, and in addition, the rationality of motor loss and other temperature simulation results can be further verified by combining an analytic model such as a motor thermal resistance model.
S3, constructing a quick prediction model of the multidimensional failure associated load of the electric drive system based on fusion of a Genetic Algorithm (GA) optimized back propagation neural network (BP) and a long-term memory neural network (LSTM) algorithm by adopting a model and data hybrid driving method; because the electric drive system bears multisource loads in the operation process, different loads have different time scale characteristics, and multiscale load prediction is difficult to realize through a single neural network model. Therefore, for signals with stress, loss and the like in quick response in an electric drive system, a mapping relation with strong correlation among the signals can be established through a GA-BP neural network optimized by a genetic algorithm. For a hysteretic time delay temperature signal, an LSTM neural network can be adopted to solve the problem of long sequence dependence, and the rapid prediction of the time delay temperature load is realized.
Specifically, in this embodiment, the algorithm described in S3 is integrated to use the predicted power loss as the main input signal of the temperature prediction model based on the GA-BP algorithm, construct a neural network intermediate layer by combining the original input layer signal and the predicted power loss in the GA-BP algorithm, and use the neural network intermediate layer as the input layer of the LSTM neural network, so as to obtain a multi-dimensional load fast prediction model of the electric drive system, thereby realizing fast prediction of different temperature loads.
Fig. 2 is a schematic diagram of a multidimensional load prediction model based on GA-BP-LSTM algorithm fusion, input load data in fig. 2 are motor rotating speed, motor torque and motor current under user working conditions, stress and loss load can be rapidly predicted through a GA-BP neural network, and predicted loss is used as an intermediate layer input signal for predicting stator winding temperature, stator core temperature, rotor core temperature and IGBT node temperature. The structure and parameters of the BP neural network are optimized, different hidden layers and different hidden layer node numbers of the BP neural network are optimized respectively, and for loads with fast response such as stress, loss and the like in an electric drive system, the prediction precision is high when a single hidden layer is adopted and the hidden layer node number is 8, as shown in figure 3. The initial threshold and the weight of the BP neural network are optimized through a genetic algorithm, and the cross probability and the variation probability are optimal parameters under the condition that the root mean square error is minimum, as shown in figure 4.
And S4, based on the electric drive system multi-dimensional failure associated load rapid prediction model fused by the neural network algorithm constructed in the S3, failure associated load prediction of key stress, power loss, time delay temperature and the like is realized. Constructing a load prediction precision evaluation index: the prediction accuracy of the multi-dimensional load rapid prediction model of the electric drive system is verified through the correlation coefficient, the root-mean-square error, the time domain frequency distribution and the damage recurrence ratio;
specifically, in this embodiment, a multidimensional failure associated load rapid prediction model of the electric drive system is used to obtain time-domain comparison between predicted values and expected values of different loads, as shown in fig. 5 and 6. Root mean square error RMSE, correlation coefficient R, described in S4 2 The expression is as follows:
in the formula, y i 、
Respectively as a target value and a predicted value;
is the average of the target values; n is the number of samples;
the time domain frequency distribution is a frequency distribution histogram of predicted values and expected values of different failure loads, and as shown in fig. 7, the distribution range of the predicted values and the expected values is substantially consistent with the frequency of each interval. The damage recurrence ratio in S4 is a degree to which the damage of each component under the expected load can be replicated in predicting the damage of each component under the load, as shown in fig. 8. The damage recurrence ratio calculation formula is as follows:
in the formula, delta i As the damage reproduction ratio of the component i,
for the damage value of component i under the predicted load,
is the damage value of component i at the desired target load.
Constructing a reasonable load prediction precision evaluation index, wherein the specific evaluation index at least comprises the following steps: the correlation coefficient of the predicted load and the target load is more than 0.9, the total frequency difference in time domain distribution is less than 10%, and the damage recurrence ratio of each component is between 0.8 and 1.2. If the index is not met, the structure and parameters of the prediction model are further optimized and adjusted.
The foregoing illustrates and describes the principles, general features, and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (9)
1. The electric drive system multidimensional load prediction method based on model and data hybrid driving is characterized by comprising the following steps:
acquiring actual user operation data, and analyzing failure mechanisms and failure associated loads of different components of the electric drive system based on the actual user operation data;
extracting a multidimensional failure associated load time domain signal by adopting a model driving method;
constructing a multi-dimensional failure associated load rapid prediction model of the electric drive system based on neural network algorithm fusion by adopting a model and data hybrid drive method;
and constructing a load prediction precision evaluation index, and verifying the effectiveness of the quick prediction model of the multi-dimensional failure associated load of the electric drive system.
2. The method of claim 1, wherein the model-based hybrid data drive electric drive system multi-dimensional load prediction is based on a model-based hybrid data drive,
the actual user operation data comprises but is not limited to vehicle speed, motor rotating speed, motor torque, motor current and motor voltage which are acquired by the electric drive system in the actual operation process;
the different components of the electric drive system include, but are not limited to, a motor shaft, gears, bearings, stator windings, rotor magnetic bridges, rotor magnetic steels, power devices IGBT.
3. The method of claim 1, wherein the model and data hybrid drive-based electric drive system multi-dimensional load prediction method,
the failure mechanism is a failure mode and a failure physical model of the different components of the electric drive system under multi-dimensional excitation load;
the failure modes include, but are not limited to, fatigue failure, wear failure, aging failure, thermomechanical fatigue failure;
the physical model of failure includes, but is not limited to, the Basquin model under the fatigue failure, the Archard wear model under the wear failure, the Arrhenius degradation model under the aging failure, the Coffin-Manson model under the thermomechanical fatigue failure;
the failure-related load is the relevant load that causes the component to fail, including but not limited to motor speed, motor torque, motor current, motor voltage, contact stress, power loss, temperature.
4. The method of claim 1, wherein the model-based hybrid data drive electric drive system multi-dimensional load prediction is based on a model-based hybrid data drive,
the model driving method includes, but is not limited to, a load acquisition method based on physical model simulation and a load acquisition method based on analytical model calculation.
5. The model and data hybrid drive-based electric drive system multi-dimensional load prediction method of claim 1, wherein the model and data hybrid drive method comprises:
and on the basis of the multi-dimensional failure associated load acquired by the model driving method, constructing a multi-dimensional load rapid prediction model of the electric drive system based on algorithm fusion of a GA (genetic algorithm) optimized BP (Back propagation) neural network and an LSTM (least Square TM) neural network by adopting a data driving method, and optimizing the structure and parameters of the multi-dimensional load rapid prediction model of the electric drive system.
6. The model and data hybrid drive-based electric drive system multi-dimensional load prediction method of claim 5, wherein the algorithm is fused as:
and on the basis of a GA-BP algorithm, taking the predicted power loss as a main input signal of a temperature prediction model, and combining an original input layer signal in the GA-BP algorithm and the predicted power loss to construct a neural network intermediate layer, wherein the neural network intermediate layer is used as an input layer of an LSTM neural network, and a multi-dimensional load rapid prediction model of the electric drive system is established.
7. The model-and-data-hybrid-drive-based multi-dimensional load prediction method for an electric drive system according to claim 5, wherein the optimization includes, but is not limited to, hidden layer number optimization of the BP neural network, hidden layer node number optimization, initial weight and threshold optimization, and initial parameter optimization of the LSTM neural network.
8. The multidimensional load prediction method for the electric drive system based on model and data hybrid driving of claim 1, wherein the load prediction accuracy evaluation index comprises but is not limited to correlation coefficient, root mean square error, time-domain frequency distribution and damage recurrence ratio;
the correlation coefficient is larger than 0.9, the difference of the time domain frequency distribution is smaller than 10%, and the damage recurrence ratio is between 0.8 and 1.2.
9. The method of claim 8, wherein the model and data hybrid drive-based electric drive system multi-dimensional load prediction method,
the root mean square error is calculated by the following formula:
the calculation formula of the correlation coefficient is as follows:
wherein, RMSE is root mean square error; r is 2 Is a correlation coefficient; y is i 、
Respectively a target value and a predicted value;
is the average of the target values; n is the number of samples;
the calculation formula of the damage recurrence ratio is as follows:
in the formula, delta i As the damage reproduction ratio of the component i,
for the damage value of component i under the predicted load,
is the damage value of component i at the desired target load.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211369253.7A CN115753455B (en) | 2022-11-03 | 2022-11-03 | Multi-dimensional load prediction method of electric drive system based on model and data hybrid drive |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211369253.7A CN115753455B (en) | 2022-11-03 | 2022-11-03 | Multi-dimensional load prediction method of electric drive system based on model and data hybrid drive |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115753455A true CN115753455A (en) | 2023-03-07 |
| CN115753455B CN115753455B (en) | 2024-02-13 |
Family
ID=85357554
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211369253.7A Active CN115753455B (en) | 2022-11-03 | 2022-11-03 | Multi-dimensional load prediction method of electric drive system based on model and data hybrid drive |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115753455B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004028948A (en) * | 2002-06-28 | 2004-01-29 | Ricoh Co Ltd | Method for deciding dynamic transmission characteristic of drive transmission system and recording medium |
| CN109687811A (en) * | 2019-01-28 | 2019-04-26 | 江苏理工学院 | The suppressing method of vehicle-mounted AC-battery power source drive system temperature effect |
| EP3663732A1 (en) * | 2018-12-05 | 2020-06-10 | Tata Consultancy Services Limited | Method and system for monitoring machine health using radar based segregation for induced machine vibrations |
| CN111859724A (en) * | 2020-05-27 | 2020-10-30 | 中铁第四勘察设计院集团有限公司 | Hybrid-driven ballastless track fatigue life prediction method and system |
| CN111931393A (en) * | 2019-04-28 | 2020-11-13 | 沈阳工业大学 | Prediction and evaluation method for residual life of key parts of waste machine tool |
| CN113092115A (en) * | 2021-04-09 | 2021-07-09 | 重庆大学 | Digital twin model construction method of digital-analog combined drive full-life rolling bearing |
| CN114881410A (en) * | 2022-04-02 | 2022-08-09 | 国网山西省电力公司电力科学研究院 | Model-data hybrid driven power system transient stability online evaluation method |
| CN114929000A (en) * | 2022-06-21 | 2022-08-19 | 温州大学 | Power supply water cooling system with mixed WBG (work breakdown voltage) and Si (silicon on insulator) devices and control strategy thereof |
| CN114936494A (en) * | 2022-05-23 | 2022-08-23 | 华东理工大学 | Data physical fusion driven high-temperature component reliability evaluation method and system |
-
2022
- 2022-11-03 CN CN202211369253.7A patent/CN115753455B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004028948A (en) * | 2002-06-28 | 2004-01-29 | Ricoh Co Ltd | Method for deciding dynamic transmission characteristic of drive transmission system and recording medium |
| EP3663732A1 (en) * | 2018-12-05 | 2020-06-10 | Tata Consultancy Services Limited | Method and system for monitoring machine health using radar based segregation for induced machine vibrations |
| CN109687811A (en) * | 2019-01-28 | 2019-04-26 | 江苏理工学院 | The suppressing method of vehicle-mounted AC-battery power source drive system temperature effect |
| CN111931393A (en) * | 2019-04-28 | 2020-11-13 | 沈阳工业大学 | Prediction and evaluation method for residual life of key parts of waste machine tool |
| CN111859724A (en) * | 2020-05-27 | 2020-10-30 | 中铁第四勘察设计院集团有限公司 | Hybrid-driven ballastless track fatigue life prediction method and system |
| CN113092115A (en) * | 2021-04-09 | 2021-07-09 | 重庆大学 | Digital twin model construction method of digital-analog combined drive full-life rolling bearing |
| CN114881410A (en) * | 2022-04-02 | 2022-08-09 | 国网山西省电力公司电力科学研究院 | Model-data hybrid driven power system transient stability online evaluation method |
| CN114936494A (en) * | 2022-05-23 | 2022-08-23 | 华东理工大学 | Data physical fusion driven high-temperature component reliability evaluation method and system |
| CN114929000A (en) * | 2022-06-21 | 2022-08-19 | 温州大学 | Power supply water cooling system with mixed WBG (work breakdown voltage) and Si (silicon on insulator) devices and control strategy thereof |
Non-Patent Citations (1)
| Title |
|---|
| 吕力行等: "基于数据 -模型混合驱动的锂电池储能系统状态估计及预警方法", 《热力发电》, vol. 50, no. 8, pages 64 - 72 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115753455B (en) | 2024-02-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ma et al. | Discriminative deep belief networks with ant colony optimization for health status assessment of machine | |
| CN110866314B (en) | Method for predicting residual life of rotating machinery of multilayer bidirectional gate control circulation unit network | |
| Yan et al. | A dynamic multi-scale Markov model based methodology for remaining life prediction | |
| CN110531266B (en) | Synchronous motor excitation winding turn-to-turn short circuit fault early warning method | |
| Li et al. | Optimal Bayesian control policy for gear shaft fault detection using hidden semi-Markov model | |
| Kielb et al. | Experimental study of aerodynamic and structural damping in a full-scale rotating turbine | |
| Li et al. | Residual useful life estimation by a data‐driven similarity‐based approach | |
| Prakash et al. | Toward a big data-based approach: A review on degradation models for prognosis of critical infrastructure | |
| Qian et al. | Structural fatigue reliability analysis based on active learning Kriging model | |
| Wen et al. | A novel machine learning model for safety risk analysis in flywheel-battery hybrid energy storage system | |
| Xia | Forecasting method for product reliability along with performance data | |
| Parvaresh et al. | Model predictive control of a hydraulic actuator in torque applying system of a mechanically closed-loop test rig for the helicopter gearbox | |
| Kelly et al. | Data-driven approach for identifying mistuning in As-Manufactured Blisks | |
| Ekwaro-Osire et al. | Incipient Fault Point Detection Based on Multiscale Diversity Entropy | |
| Bhatti et al. | Intelligent fault diagnosis mechanism for industrial robot actuators using digital twin technology | |
| CN115753455A (en) | Electric drive system multidimensional load prediction method based on model and data hybrid driving | |
| Melendez et al. | Self-supervised Multi-stage Estimation of Remaining Useful Life for Electric Drive Units | |
| Bradley et al. | A general approach for current-based condition monitoring of induction motors | |
| Xiong et al. | Digital twin-based degradation prediction for train electro-pneumatic valve | |
| Shanling et al. | Real-time rubber quality model based on CNN-LSTM deep learning theory | |
| CN114741922B (en) | Turbine blade creep-fatigue life prediction method based on Attention mechanism | |
| Chatterton et al. | An unconventional method for the diagnosis and study of generator rotor thermal bows | |
| Zhou et al. | A fatigue life prediction method for the drive system of wind turbine using internet of things | |
| CN114936494A (en) | Data physical fusion driven high-temperature component reliability evaluation method and system | |
| Wu et al. | Multi-level health degree analysis of vehicle transmission system based on PSO-BP neural network data fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |