CN115085627A - Motor parameter dynamic identification method - Google Patents
Motor parameter dynamic identification method Download PDFInfo
- Publication number
- CN115085627A CN115085627A CN202211003896.XA CN202211003896A CN115085627A CN 115085627 A CN115085627 A CN 115085627A CN 202211003896 A CN202211003896 A CN 202211003896A CN 115085627 A CN115085627 A CN 115085627A
- Authority
- CN
- China
- Prior art keywords
- motor
- neural network
- parameters
- parameter
- model
- 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.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P23/00—Arrangements or methods for the control of AC motors characterised by a control method other than vector control
- H02P23/14—Estimation or adaptation of motor parameters, e.g. rotor time constant, flux, speed, current or voltage
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
- H02P23/00—Arrangements or methods for the control of AC motors characterised by a control method other than vector control
- H02P23/0004—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control
- H02P23/0018—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control using neural networks
Abstract
The invention relates to the field of motor parameter dynamic identification, and provides a motor parameter dynamic identification method, which comprises the following steps: firstly, coupling motor parameters under the same model, wherein the motor parameters comprise a back electromotive force constant, a rotational inertia, and a resistance and an inductance of a stator; secondly, constructing and training a neural network by using coupled motor parameters based on the motor model and the motor running state, wherein the motor running state comprises no load, load and overload; and then, setting time domain characteristics for the voltage and the current of the motor, decoupling motor parameters by using the trained neural network, and dynamically identifying the motor parameters based on the motor model and the motor running state when the motor runs. According to the invention, accurate motor parameters can be dynamically identified in real time under different running states of the motor according to the model difference of the motor, and a controller is not required to be constructed, so that the dynamic identification of the motor parameters has higher accuracy and better real-time property.
Description
Technical Field
The invention relates to the field of motor parameter dynamic identification, in particular to a motor parameter dynamic identification method.
Background
At present, the identification of motor parameters is mostly realized by designing a corresponding controller, for example, patent application No. CN202210340561.0 discloses a motor parameter identification method based on a high-frequency voltage injection method considering phase resistance, which includes injecting high-frequency sinusoidal voltage into a permanent magnet synchronous motor, then collecting real-time phase current of the motor, obtaining a current signal for identification by filtering high-frequency components in the signal through coordinate transformation, mathematical calculation and first-order low-pass filtering, and finally accurately identifying the initial position of a magnetic pole, the phase resistance of a stator, a direct axis inductance and a quadrature axis inductance by using a closed-loop tracking method.
However, due to the conventional controllers such as the PID controller and the LRQ controller, the logic used in the design and the calculation amount related to the parameter setting of the open/close loop are large, and the closed-loop parameter adjustment has no specific reference index, so that the real-time performance and the accuracy of the identification of the motor parameter cannot be ensured.
Disclosure of Invention
The invention aims to provide a motor parameter dynamic identification method with high real-time performance and accuracy.
The invention solves the technical problem, and adopts the technical scheme that:
the motor parameter dynamic identification method comprises the following steps:
coupling motor parameters under the same model, wherein the motor parameters comprise a back electromotive force constant, a rotational inertia, and resistance and inductance of a stator;
constructing and training a neural network by using coupled motor parameters based on the motor model and the motor running state, wherein the motor running state comprises no load, load and overload;
time domain characteristics are set for the voltage and the current of the motor, motor parameters are decoupled by utilizing the trained neural network, and when the motor runs, the motor parameters are dynamically identified based on the motor model and the motor running state.
Further, an identification error threshold value of each motor parameter is set in the constructed neural network, if the difference value between each dynamically identified motor parameter and the corresponding measured value is within the identification error threshold value, the dynamic motor parameter output by the neural network is taken as the standard for recording, otherwise, the measured value is taken as the standard for recording, and the neural network parameter is adjusted.
Further, when the difference value between each dynamically identified motor parameter and the measured value corresponding to the dynamically identified motor parameter is within the identification error threshold value, the identification error is fed back to the neural network, the motor parameter value dynamically identified by each time node is compared and updated with the motor parameter value dynamically identified by the previous time node, and the latest dynamically identified motor parameter value is used as the final output value.
Further, when the recognition error is fed back to the neural network, the parameters of the neural network model are optimally updated.
Furthermore, when the neural network is trained by each motor parameter under the same model, the number of the motor parameters is not less than 1000.
Further, the neural network is a BP neural network.
The beneficial effects of the invention are: according to the motor parameter dynamic identification method, firstly, motor parameters under the same model are coupled, wherein the motor parameters comprise a back electromotive force constant, a rotational inertia, a resistance and an inductance of a stator and the like; secondly, constructing and training a neural network by using coupled motor parameters based on the motor model and the motor running state, wherein the motor running state comprises no load, load and overload; and then, setting time domain characteristics for the voltage and the current of the motor, decoupling motor parameters by using the trained neural network, and dynamically identifying the motor parameters based on the motor model and the motor running state when the motor runs. Therefore, the controller is not required to be designed in the application, the motor parameters are dynamically adjusted and optimized for identification, and the real-time and accurate motor parameters can be obtained only by inputting all the parameters of the motor in all the running states to the utilization neural network and carrying out real-time dynamic optimization.
Drawings
Fig. 1 is a flowchart of a motor parameter dynamic identification method according to the present invention.
Detailed Description
The technical scheme of the invention is described in detail in the following with reference to the accompanying drawings.
The invention provides a motor parameter dynamic identification method, a flow chart of which is shown in figure 1, wherein the method can comprise the following steps:
s1, coupling motor parameters under the same model, wherein the motor parameters comprise a back electromotive force constant, a rotational inertia, and a resistance and an inductance of a stator;
s2, constructing and training a neural network by using coupled motor parameters based on the motor model and the motor running state, wherein the motor running state comprises no load, load and overload;
and S3, setting time domain characteristics for the voltage and the current of the motor, decoupling motor parameters by using the trained neural network, and dynamically identifying the motor parameters based on the motor model and the motor running state when the motor runs.
In the application, what first to do is to couple each motor parameter that needs to be identified, because the current conventional mode that carries out motor parameter identification through the controller mostly needs real-time calculation, lead to whole parameter identification process to have a delay, and because the calculated amount is great, the accuracy that the calculated result probably can appear can not be guaranteed, and simultaneously, still because numerous computational formula, when involving controller parameter adjustment, lead to parameter adjustment process also comparatively loaded down with trivial details, therefore, this application does not carry out controller design, only need to couple the motor parameter that needs to be identified earlier, rethread collection sample training neural network, when motor under a certain model is under no-load, load or overload state, the dynamic identification of all parameters under this running state of homoenergetic realization this model, not only the recognition accuracy is high, and the real-time is also very high.
It should be noted that, because the motor models are various and the applied working conditions are different, the identification requirements for the motor parameters are not completely the same under many circumstances, and therefore, in the present application, when the motor parameters are dynamically identified, the motor parameters are decoupled by using the neural network, and on the premise of knowing the motor models and the motor operating state, accurate and real-time dynamic identification of each parameter can be achieved.
In the application, a neural network can be constructed to be provided with identification error threshold values of various motor parameters, the identification error threshold value can be adjusted in real time according to the operation condition of the motor, if the difference value between each motor parameter which is dynamically identified and the measured value corresponding to the motor parameter is within the identification error threshold value, recording the dynamic motor parameters output by the neural network, otherwise recording the dynamic motor parameters output by the neural network based on the measured values, and adjusting the neural network parameters, when the neural network parameters need to be adjusted, the recognition error is large, which indicates that the representativeness of the data set utilized is not particularly obvious when the application network is trained, and therefore, the neural network can be continuously optimized and trained, and the parameters of the neural network can also be directly adjusted, so that the recognition error is controllable, and the neural network is also effective within the range of the set recognition error threshold.
In the practical application process, when the difference value between each dynamically-identified motor parameter and the measured value corresponding to the dynamically-identified motor parameter is within the identification error threshold value, the identification error can be fed back to the neural network, so that the neural network can know the identification error of the current time node, a basis is made for the optimization of the subsequent neural network, the motor parameter value dynamically identified by each time node can be compared and updated with the motor parameter value dynamically identified by the previous time node, and the latest dynamically-identified motor parameter value is used as a final output value.
It should be noted that, when the identification error is fed back to the neural network, in order to ensure the accuracy of the dynamic identification of the motor parameter by the subsequent time node, the parameters of the neural network model may also be optimized and updated.
In order to ensure that the trained neural network can more accurately identify the motor parameters, the number of the training sample sets of the electromotive force constant, the rotational inertia and the resistance and inductance of the stator can be restricted, for example, the number of each motor parameter in the same model is not less than 1000 when the neural network is trained.
In the practical application process, a variety of neural networks can be constructed and trained, and considering that the BP neural network has strong nonlinear mapping capability, and the parameters of each motor are not completely linear with the parameters of the motor in the application under different operation states, therefore, the neural network in the application can be preferably the BP neural network.
Claims (6)
1. The motor parameter dynamic identification method is characterized by comprising the following steps:
coupling motor parameters under the same model, wherein the motor parameters comprise a back electromotive force constant, a rotational inertia, and resistance and inductance of a stator;
constructing and training a neural network by using coupled motor parameters based on the motor model and the motor running state, wherein the motor running state comprises no load, load and overload;
time domain characteristics are set for the voltage and the current of the motor, motor parameters are decoupled by utilizing the trained neural network, and when the motor runs, the motor parameters are dynamically identified based on the motor model and the motor running state.
2. The method according to claim 1, wherein the neural network is configured to have an identification error threshold for each motor parameter, and if the difference between each dynamically identified motor parameter and its corresponding measured value is within the identification error threshold, the dynamic motor parameter outputted by the neural network is recorded based on the identified value, otherwise, the actual measured value is recorded based on the measured value, and the neural network parameter is adjusted.
3. The method according to claim 2, wherein when the difference between each dynamically identified motor parameter and its corresponding measured value is within the threshold value of the identification error, the identification error is fed back to the neural network, the motor parameter value dynamically identified by each time node is compared and updated with the motor parameter value dynamically identified by the previous time node, and the latest dynamically identified motor parameter value is used as the final output value.
4. The dynamic motor parameter identification method of claim 3, wherein the parameters of the neural network model are optimally updated when the identification error is fed back to the neural network.
5. The method according to claim 1, wherein the number of each motor parameter in the same model is not less than 1000 when training the neural network.
6. The method according to any of claims 1-5, wherein the neural network is a BP neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211003896.XA CN115085627A (en) | 2022-08-22 | 2022-08-22 | Motor parameter dynamic identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211003896.XA CN115085627A (en) | 2022-08-22 | 2022-08-22 | Motor parameter dynamic identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115085627A true CN115085627A (en) | 2022-09-20 |
Family
ID=83244277
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211003896.XA Pending CN115085627A (en) | 2022-08-22 | 2022-08-22 | Motor parameter dynamic identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115085627A (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102497156A (en) * | 2011-12-27 | 2012-06-13 | 东南大学 | Neural-network self-correcting control method of permanent magnet synchronous motor speed loop |
| CN102937670A (en) * | 2012-11-17 | 2013-02-20 | 河南科技大学 | Method for identifying resistance parameters of rotors of induction motor on basis of Elman neural network |
| CN103248306A (en) * | 2013-05-24 | 2013-08-14 | 天津大学 | Online decoupling identification method of multiple parameters of PMSM (permanent magnet synchronous motor) |
| CN103793746A (en) * | 2014-02-19 | 2014-05-14 | 上海交通大学 | Method for identifying parameters of coal-fired power plant boiler superheater model |
| CN104767449A (en) * | 2015-03-02 | 2015-07-08 | 江苏大学 | Bearing-free asynchronous motor RBF neural network self-adaptive inverse decoupling control and parameter identification method |
| CN105588930A (en) * | 2015-12-17 | 2016-05-18 | 镇江市高等专科学校 | Method for measuring parameters of soil in greenhouse |
| CN110535396A (en) * | 2019-08-30 | 2019-12-03 | 长安大学 | Surface permanent magnetic Synchronous Machine Models forecast Control Algorithm based on BP neural network |
| CN112152529A (en) * | 2020-09-28 | 2020-12-29 | 长沙贝士德电气科技有限公司 | Maximum thrust control method and system for permanent magnet linear motor |
| CN112152539A (en) * | 2020-09-29 | 2020-12-29 | 中国船舶重工集团公司第七二四研究所 | Neural network compensation motor load moment observer implementation method |
| CN112622645A (en) * | 2021-03-09 | 2021-04-09 | 成都微精电机股份公司 | Self-adjusting method for fully-automatic control motor of vehicle |
| CN114244216A (en) * | 2021-12-17 | 2022-03-25 | 武汉华中天勤防务技术有限公司 | Permanent magnet synchronous motor parameter identification method, device and system |
| CN114492644A (en) * | 2022-01-28 | 2022-05-13 | 上海电机学院 | Motor fault detection method based on improved neural network |
| CN114572053A (en) * | 2022-03-04 | 2022-06-03 | 中南大学 | Electric automobile energy management method and system based on working condition identification |
| CN114915225A (en) * | 2022-04-22 | 2022-08-16 | 昆明理工大学 | Permanent magnet synchronous motor parameter identification method based on optimized BP neural network |
-
2022
- 2022-08-22 CN CN202211003896.XA patent/CN115085627A/en active Pending
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102497156A (en) * | 2011-12-27 | 2012-06-13 | 东南大学 | Neural-network self-correcting control method of permanent magnet synchronous motor speed loop |
| CN102937670A (en) * | 2012-11-17 | 2013-02-20 | 河南科技大学 | Method for identifying resistance parameters of rotors of induction motor on basis of Elman neural network |
| CN103248306A (en) * | 2013-05-24 | 2013-08-14 | 天津大学 | Online decoupling identification method of multiple parameters of PMSM (permanent magnet synchronous motor) |
| CN103793746A (en) * | 2014-02-19 | 2014-05-14 | 上海交通大学 | Method for identifying parameters of coal-fired power plant boiler superheater model |
| CN104767449A (en) * | 2015-03-02 | 2015-07-08 | 江苏大学 | Bearing-free asynchronous motor RBF neural network self-adaptive inverse decoupling control and parameter identification method |
| CN105588930A (en) * | 2015-12-17 | 2016-05-18 | 镇江市高等专科学校 | Method for measuring parameters of soil in greenhouse |
| CN110535396A (en) * | 2019-08-30 | 2019-12-03 | 长安大学 | Surface permanent magnetic Synchronous Machine Models forecast Control Algorithm based on BP neural network |
| CN112152529A (en) * | 2020-09-28 | 2020-12-29 | 长沙贝士德电气科技有限公司 | Maximum thrust control method and system for permanent magnet linear motor |
| CN112152539A (en) * | 2020-09-29 | 2020-12-29 | 中国船舶重工集团公司第七二四研究所 | Neural network compensation motor load moment observer implementation method |
| CN112622645A (en) * | 2021-03-09 | 2021-04-09 | 成都微精电机股份公司 | Self-adjusting method for fully-automatic control motor of vehicle |
| CN114244216A (en) * | 2021-12-17 | 2022-03-25 | 武汉华中天勤防务技术有限公司 | Permanent magnet synchronous motor parameter identification method, device and system |
| CN114492644A (en) * | 2022-01-28 | 2022-05-13 | 上海电机学院 | Motor fault detection method based on improved neural network |
| CN114572053A (en) * | 2022-03-04 | 2022-06-03 | 中南大学 | Electric automobile energy management method and system based on working condition identification |
| CN114915225A (en) * | 2022-04-22 | 2022-08-16 | 昆明理工大学 | Permanent magnet synchronous motor parameter identification method based on optimized BP neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103427746B (en) | A kind of face mounted permagnetic synchronous motor full speed range method for controlling position-less sensor | |
| CN106788044B (en) | A kind of adaptive non-singular terminal sliding-mode control of permanent magnet synchronous motor based on interference observer | |
| CN104953913B (en) | Networking alternating current generator LS-SVM generalized inverse decoupling control methods based on active disturbance rejection | |
| CN101917151B (en) | Method for controlling position-less sensor of switched reluctance motor | |
| CN111600518A (en) | Design method of permanent magnet synchronous current controller based on extended state observer | |
| CN109672383A (en) | A kind of salient-pole permanent-magnet synchronous motor on-line parameter discrimination method | |
| CN111942156A (en) | Permanent magnet synchronous motor demagnetization fault detection method and device and automobile | |
| CN110932642B (en) | Hermite interpolation-based transient phase torque estimation method for switched reluctance motor | |
| CN112054730A (en) | Permanent magnet synchronous motor online parameter identification method | |
| CN105024612A (en) | Parameter identification-based motor current control method and system | |
| CN114944801A (en) | PMSM (permanent magnet synchronous motor) position sensorless control method based on innovation self-adaptive extended Kalman | |
| CN109510539B (en) | Model prediction flux linkage control system and method based on gain matrix | |
| Song et al. | A comparative study on modeling methods for switched reluctance machines | |
| Wang et al. | Improved auto disturbance rejection control based on moth flame optimization for permanent magnet synchronous motor | |
| CN108565896B (en) | Photovoltaic power generation system stability analysis method considering power control influence | |
| CN107834713B (en) | Tuning feedback control method of wireless power transmission system considering dynamic performance | |
| CN115085627A (en) | Motor parameter dynamic identification method | |
| CN108574440A (en) | A kind of permanent magnet synchronous motor method for estimating state based on sliding formwork reference adaptive | |
| CN109495040B (en) | Motor PI control parameter setting method based on dynamic inductor | |
| CN114785216B (en) | Design method of self-adaptive gain sliding mode observer | |
| Pengcheng et al. | Offline parameter identification strategy of permanent magnet synchronous motor considering the inverter nonlinearities | |
| CN112152538A (en) | Inverse model control method of permanent magnet synchronous motor | |
| CN112487748B (en) | Lithium ion battery fractional order model establishment method considering time-frequency domain characteristics | |
| CN110048659B (en) | Motor control optimization system and method | |
| CN113067506A (en) | Permanent magnet synchronous motor periodic disturbance suppression method based on inner model equivalent input interference |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220920 |
|
| RJ01 | Rejection of invention patent application after publication |