A control method and system for an embedded permanent magnet synchronous motor

Abstract

The present invention relates to the field of motor control technology, and specifically, to a control method and system for an embedded permanent magnet synchronous motor. The present invention can accurately construct motor digital-analog data by obtaining motor variable data to determine synchronous rotation coordinate transformation information, laying a solid foundation for subsequent control. The first control strategy information and component characteristics and the second control strategy information are generated based on the motor variable data, and the control component weights can be carefully weighed to make the torque control decision parameters more in line with the actual needs of the motor. Finally, control processing is performed based on this information, which can achieve accurate and efficient control of the embedded permanent magnet synchronous motor. This control method helps to improve the operating efficiency of the motor, optimize torque output, enhance the adaptability of the motor to different working conditions, reduce energy consumption and extend the service life of the motor, and improve the overall performance of the motor.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and in particular to a control method and system for an embedded permanent magnet synchronous motor. Background Art

[0002] In the existing field of embedded permanent magnet synchronous motor control technology, there are many problems that need to be solved. Traditional control methods often find it difficult to accurately and effectively control the motor according to its actual operating state. On the one hand, there is a lack of comprehensive and in-depth analysis and utilization of motor variable data, and it is impossible to accurately construct digital-analog data that reflects the real state of the motor, which leads to a lack of precise basis for control decisions. On the other hand, the formulation of control strategies is not detailed and flexible enough, and it is impossible to reasonably weigh the weights of different control components, which makes torque control not accurate enough and it is difficult for torque output to reach the optimal state. At the same time, it cannot adapt well to the operating requirements of the motor under different working conditions, which can easily cause problems such as energy waste, low motor efficiency and shortened service life. Summary of the invention

[0003] In order to at least overcome the above-mentioned deficiencies in the prior art, one of the objectives of the present invention is to provide a control method and system for an embedded permanent magnet synchronous motor.

[0004] An embodiment of the present invention provides a control method for an embedded permanent magnet synchronous motor, the method being applied to an embedded permanent magnet synchronous motor control system, the method comprising:

[0005] Acquire motor variable data for making vector control decisions, and determine synchronous rotation coordinate transformation information from the motor variable data, wherein the synchronous rotation coordinate transformation information represents motor digital-analog data corresponding to the embedded permanent magnet synchronous motor to be controlled;

[0006] generating first control strategy information and component characteristics of the synchronous rotating coordinate transformation information according to the motor variable data, and generating second control strategy information according to the synchronous rotating coordinate transformation information, wherein the component characteristics indicate the control component weight of the synchronous rotating coordinate transformation information relative to the first control strategy information, and the second control strategy information includes a torque control decision parameter corresponding to the synchronous rotating coordinate transformation information;

[0007] The embedded permanent magnet synchronous motor to be controlled is controlled according to the first control strategy information and the component characteristics, and the second control strategy information.

[0008] Preferably, determining the synchronous rotation coordinate transformation information from the motor variable data comprises:

[0009] Derived motor variable data is generated through the motor variable data, wherein the derived motor variable data includes the synchronous rotating coordinate transformation information, magnetic field component information, and timing control characteristics of the embedded permanent magnet synchronous motor to be controlled based on the synchronous rotating coordinate transformation information and the magnetic field component information, and the magnetic field component information is used to optimize the synchronous rotating coordinate transformation information.

[0010] Preferably, generating derived motor variable data through the motor variable data includes: preprocessing the motor variable data using a decision control network to determine the synchronous rotation coordinate transformation information, the magnetic field component information and the derived motor variable data from the motor variable data.

[0011] Preferably, the first control strategy information and component characteristics for generating the synchronous rotation coordinate transformation information according to the motor variable data include:

[0012] Using a long short-term memory network, generating the first control strategy information according to the motor variable data;

[0013] Determining weight semantics of the first control strategy information using the long short-term memory network according to the generated first control strategy information, wherein the weight semantics indicates a reference confidence of the synchronous rotation coordinate transformation information in the first control strategy information;

[0014] The weight semantics are adjusted to generate component features of the synchronous rotation coordinate transformation information.

[0015] Preferably, the method further comprises: determining discrete state prediction information based on the second control strategy information, wherein the discrete state prediction information indicates a discrete state trend of the torque control decision parameter compared to the second control strategy information.

[0016] Preferably, the control processing of the embedded permanent magnet synchronous motor to be controlled according to the first control strategy information and the component characteristics, and the second control strategy information includes:

[0017] extracting the torque control decision parameter from the second control strategy information according to the discrete state prediction information;

[0018] A control execution task label is generated according to the component characteristics, the first control strategy information and the torque control decision parameter to perform control processing on the embedded permanent magnet synchronous motor to be controlled.

[0019] Preferably, generating a control execution task label according to the component characteristics, the first control strategy information and the torque control decision parameter to control the embedded permanent magnet synchronous motor to be controlled includes:

[0020] Based on three dimensional descriptions corresponding to the control dimension characteristics of the first control strategy information, the control dimension characteristics of the second control strategy information, and the linkage dimension characteristics between the second control strategy information and the first control strategy information, the embedded permanent magnet synchronous motor to be controlled is controlled through multiple task node labels.

[0021] Preferably, controlling the embedded permanent magnet synchronous motor to be controlled by using multiple task node tags includes:

[0022] In each parameter control cycle corresponding to each task node label, according to the dimensional description corresponding to the task node label in the three dimensional descriptions, vector control based on coordinate transformation is performed on the embedded permanent magnet synchronous motor to be controlled;

[0023] For the first task node tag among the multiple task node tags, the dimensional description corresponding to the first task node tag is the three dimensional descriptions; for the second task node tag among the multiple task node tags, the dimensional description corresponding to the second task node tag is the dimensional description corresponding to the control dimensional characteristics of the first control strategy information, and the dimensional description corresponding to the control dimensional characteristics of the second control strategy information; for the third task node tag among the multiple task node tags, the dimensional description corresponding to the third task node tag is the dimensional description corresponding to the control dimensional characteristics of the second control strategy information.

[0024] Preferably, the method further comprises:

[0025] In response to a control adjustment instruction for the embedded permanent magnet synchronous motor to be controlled, performing sensorless control processing on the embedded permanent magnet synchronous motor to be controlled;

[0026] Wherein, the sensorless control process includes:

[0027] Obtain an electromotive force statistical data sequence of the embedded permanent magnet synchronous motor to be controlled, wherein the electromotive force statistical data sequence includes X consecutive groups of electromotive force statistical data, and X is an integer greater than or equal to 1; obtain a loss statistical data sequence according to the electromotive force statistical data sequence, wherein the loss statistical data sequence includes X consecutive groups of loss statistical data; based on the electromotive force statistical data sequence, obtain an electromotive force vector distribution sequence through a first vector mining branch included in an extended Kalman filter network, wherein the electromotive force vector distribution sequence includes X electromotive force vector distributions; based on the loss statistical data sequence, obtain a loss vector distribution sequence through a second vector mining branch included in the extended Kalman filter network, wherein the loss vector distribution sequence includes X loss vector distributions; based on the electromotive force vector distribution sequence and the loss vector distribution sequence, obtain a rule discrimination weight corresponding to the electromotive force statistical data through a control output branch included in the extended Kalman filter network; determine a sensorless control mechanism for the embedded permanent magnet synchronous motor to be controlled according to the rule discrimination weight.

[0028] An embodiment of the present invention also provides an embedded permanent magnet synchronous motor control system, comprising a processor and a memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the above-mentioned embedded permanent magnet synchronous motor control method.

[0029] An embodiment of the present invention further provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the control method of the embedded permanent magnet synchronous motor is implemented.

[0030] A control method and system for an embedded permanent magnet synchronous motor provided by an embodiment of the present invention can accurately construct the motor digital-analog data by obtaining the motor variable data to determine the synchronous rotation coordinate transformation information, laying a solid foundation for subsequent control. The first control strategy information and component characteristics and the second control strategy information are generated based on the motor variable data, and the control component weights can be carefully weighed to make the torque control decision parameters more in line with the actual needs of the motor. Finally, control processing is performed based on this information to achieve accurate and efficient control of the embedded permanent magnet synchronous motor. This control method helps to improve the operating efficiency of the motor, optimize torque output, enhance the adaptability of the motor to different working conditions, reduce energy consumption and extend the service life of the motor, and improve the overall performance of the motor. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 The present invention provides a flow chart of a method for controlling an embedded permanent magnet synchronous motor.

[0033] Figure 2 A block diagram of an embedded permanent magnet synchronous motor control system provided by an embodiment of the present invention.

[0034] icon:

[0035] 100-Embedded permanent magnet synchronous motor control system;

[0036] 101 - processor; 102 - memory; 103 - bus. DETAILED DESCRIPTION

[0037] The exemplary embodiments disclosed in the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0038] In order to better understand the above technical scheme, the technical scheme of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the specific features of the present invention and the embodiments are detailed descriptions of the technical scheme of the present invention, rather than limitations on the technical scheme of the present invention. In the absence of conflict, the technical features of the present invention and the embodiments can be combined with each other.

[0039] Figure 1 This is a flow chart of a control method for an embedded permanent magnet synchronous motor according to an embodiment of the present invention, which is applied to an embedded permanent magnet synchronous motor control system and includes S110 - S130 .

[0040] S110: Acquire motor variable data for making vector control decisions, and determine synchronous rotation coordinate transformation information from the motor variable data, wherein the synchronous rotation coordinate transformation information represents motor digital-to-analog data corresponding to the embedded permanent magnet synchronous motor to be controlled.

[0041] S120: Generate first control strategy information and component characteristics of the synchronous rotation coordinate transformation information based on the motor variable data, and generate second control strategy information based on the synchronous rotation coordinate transformation information, wherein the component characteristics indicate the control component weight of the synchronous rotation coordinate transformation information relative to the first control strategy information, and the second control strategy information includes torque control decision parameters corresponding to the synchronous rotation coordinate transformation information.

[0042] S130: Control the embedded permanent magnet synchronous motor to be controlled according to the first control strategy information and the component characteristics, and the second control strategy information.

[0043] In the embedded permanent magnet synchronous motor control system, the implementation process of the entire application is a multi-link, interrelated and complex operation process. In S110, obtaining the motor variable data for vector control decision-making is the starting point for the construction of the entire control system. The motor variable data contains many key elements that can reflect the actual operating status of the embedded permanent magnet synchronous motor.

[0044] Let's first look at the acquisition of stator current data, which is accomplished by installing high-precision current sensors at specific locations on the motor's stator windings. For example, for an embedded permanent magnet synchronous motor with a rated power of 5 kilowatts, its stator winding adopts a three-phase structure. The current sensor installed on each phase winding can accurately measure the magnitude and direction of the current, with a measurement accuracy of ±0.1 amperes. The stator voltage data is obtained from the power supply terminal connected to the motor and the voltage detection device in the internal circuit of the motor. These voltage detection devices are distributed at key nodes of the circuit to ensure that the motor input voltage and the voltage drop caused by factors such as winding inductance and resistance inside the motor can be accurately obtained. For example, for a motor system with a rated voltage of 380 volts, the resolution of the voltage detection device can reach 1 volt, which can accurately reflect subtle changes in voltage.

[0045] Motor speed data is also important motor variable data. There are many ways to obtain motor speed data. One is to directly use a speed sensor installed on the motor shaft. This speed sensor can be photoelectric or magnetoelectric. Taking the photoelectric speed sensor as an example, a disk with evenly distributed slits is installed on the motor shaft. When the motor rotates, light is received by the photoelectric element through the slits. The motor speed can be obtained by calculating the number of times the light is blocked per unit time. Another method is to estimate based on the motor back electromotive force. Since the motor generates back electromotive force when it rotates, its frequency is related to the motor speed. By detecting the frequency of the back electromotive force and combining it with known parameters such as the number of pole pairs of the motor, the motor speed can be estimated. For a four-pole embedded permanent magnet synchronous motor, when the frequency of the back electromotive force is 50 Hz, based on the relationship between the motor speed, frequency and pole pair number (motor speed = 120 × frequency ÷ pole pair number), the motor speed can be calculated to be 1500 revolutions per minute.

[0046] The synchronous rotating coordinate transformation information is determined from these rich motor variable data. This information represents the motor digital-analog data corresponding to the embedded permanent magnet synchronous motor to be controlled. Synchronous rotating coordinate transformation is the core concept of vector control. Its implementation requires complex mathematical transformation of the variables in the three-phase stationary coordinate system of the motor. In the three-phase stationary coordinate system, the relationship between the motor's current, voltage and other variables is relatively complex. By using synchronous rotating coordinate transformation, these variables are converted to a synchronous rotating coordinate system oriented by the motor's rotor magnetic field. The stator current can be decomposed into the current component that generates torque (usually called torque current (i_{q})) and the current component that generates the magnetic field (usually called excitation current (i_{d})), which greatly simplifies the mathematical model of the motor and facilitates subsequent control decisions.

[0047] Enter S120, and generate the first control strategy information and component characteristics of the synchronous rotation coordinate transformation information based on the motor variable data that has been acquired. The generation of the first control strategy information is a process based on comprehensive consideration of the motor variable data. For example, when the motor variable data shows that the motor is in a full-load operation state and the temperature of the motor has an upward trend, the first control strategy information needs to take into account factors such as the heat dissipation and overload protection of the motor. If the heat dissipation capacity of the motor is limited, the first control strategy information may tend to limit the output power of the motor to prevent the motor from overheating and damage.

[0048] Take a specific numerical example to illustrate the component characteristics. For example, at a certain operating moment, the motor needs to output a larger torque to overcome the load resistance. At this time, after analyzing and calculating the motor variable data, it is determined that the torque component weight in the synchronous rotating coordinate transformation information is 0.7, and the magnetic field component weight is 0.3. This means that in the first control strategy information, the importance of torque control is higher than that of magnetic field control. This weight distribution is determined based on the current operating requirements of the motor and the performance characteristics of the motor itself. For example, for an embedded permanent magnet synchronous motor, since its permanent magnet provides a certain magnetic field, in some cases the weight of the magnetic field component can be appropriately reduced, and more attention can be paid to the control of the torque component to improve the output torque efficiency of the motor.

[0049] At the same time, the second control strategy information is generated according to the synchronous rotating coordinate transformation information, wherein the second control strategy information includes a torque control decision parameter corresponding to the synchronous rotating coordinate transformation information. This torque control decision parameter is a key factor directly related to the motor torque output control. For example, when the motor needs to achieve rapid acceleration in a short time, the torque control decision parameter may be set to a larger value based on factors such as the rated torque, current speed, load conditions, and target acceleration time of the motor. Taking an embedded permanent magnet synchronous motor with a rated torque of 100 Nm as an example, if it is to accelerate from standstill to 1000 rpm within 2 seconds, after considering factors such as the motor's moment of inertia and load torque, the torque control decision parameter may be set to 120 Nm to ensure that the expected acceleration effect can be achieved.

[0050] Finally, in step S130, the embedded permanent magnet synchronous motor to be controlled is controlled based on the first control strategy information and component characteristics, as well as the second control strategy information. This process is a key link in the implementation of the entire application, and requires the various control information generated in the previous two steps to be organically integrated and applied to the actual control of the motor.

[0051] In actual control operations, the first control strategy information and component characteristics and the second control strategy information cooperate and influence each other. For example, when the first control strategy information emphasizes the need to balance torque output and magnetic field control during medium-speed operation (according to the component characteristics, the torque component weight is 0.5 and the magnetic field component weight is also 0.5), and the torque control decision parameters in the second control strategy information show that a moderate torque output needs to be maintained at this time, such as 50 Nm. In this case, the control system will adjust the direct-axis component (related to magnetic field control) and the quadrature-axis component (related to torque control) of the stator current according to the mathematical model of the motor and the current motor variable data. For this embedded permanent magnet synchronous motor with a rated power of 5 kilowatts, through calculation and adjustment, the direct-axis current component may be set to 3 amperes and the quadrature-axis current component may be set to 4 amperes to achieve the desired torque output and magnetic field strength.

[0052] Taking into account the special properties of the embedded permanent magnet synchronous motor, such as its salient pole effect (the direct-axis inductance and the quadrature-axis inductance are not equal), special treatment is required in the entire control process. When generating the first control strategy information and the second control strategy information, the current component must be reasonably distributed according to the salient pole effect of the motor to achieve the best control effect. For example, for an embedded permanent magnet synchronous motor with a direct-axis inductance of 0.05 henry and a quadrature-axis inductance of 0.1 henry, when running at high speed and requiring a large torque output, the quadrature-axis inductance is large due to the existence of the salient pole effect, which means that under the same current change, the flux change in the quadrature-axis direction is relatively small. Therefore, when generating the first control strategy information, the weight of the quadrature-axis current component may be appropriately increased to make full use of the torque characteristics of the motor and improve the output torque of the motor.

[0053] In the process of processing synchronous rotating coordinate transformation information, its accuracy is crucial to the entire control process. When determining this information from the motor variable data, it is necessary to use precise algorithms and have powerful data processing capabilities. For example, when converting three-phase current and voltage data into direct and quadrature axis components in a synchronous rotating coordinate system, complex coordinate transformation calculations are required. This involves mathematical operations such as trigonometric operations and matrix transformations. If there are calculation errors or improper data processing in this process, it may lead to inaccurate synchronous rotating coordinate transformation information, which in turn affects the generation of the first control strategy information and the second control strategy information, and ultimately affects the control effect of the motor.

[0054] From the perspective of the overall operation process, the implementation process of this application fully considers the various characteristics of the embedded permanent magnet synchronous motor by obtaining motor variable data, generating different control strategy information and controlling the motor based on this information, and establishes close links between various links to achieve efficient and precise control of the motor.

[0055] During the startup phase of the motor, the control system determines the initial synchronous rotation coordinate transformation information based on the motor variable data at startup. For example, when the motor is just starting, the stator current gradually increases from zero, and the motor speed is zero. Based on these initial motor variable data, the synchronous rotation coordinate transformation information is determined by a specific algorithm. Then, the first control strategy information and the second control strategy information suitable for startup are generated based on factors such as the motor load condition and the startup time requirements. If the load is light when the motor starts and a quick startup is required, the first control strategy information may set the torque component weight higher, and the torque control decision parameter in the second control strategy information will also be set to a larger value, so that the motor can start smoothly and quickly.

[0056] During the acceleration process of the motor, as the motor speed increases, the motor variable data continues to change. For example, the magnitude and frequency of the stator current will change as the speed increases, and the motor's back electromotive force will increase accordingly. The control system will obtain these changing motor variable data in real time, re-execute the S110-S130 process, and adjust the control strategy to meet the requirements for torque and magnetic field during acceleration. In this process, if the acceleration slope of the motor is large, which means that a larger torque output is required, the control system will re-determine the synchronous rotation coordinate transformation information based on the new motor variable data, adjust the component characteristics in the first control strategy information, increase the weight of the torque component, and at the same time increase the torque control decision parameters in the second control strategy information to ensure that the motor can accelerate according to the expected acceleration curve.

[0057] During the steady-state operation of the motor, the control strategy also needs to be continuously adjusted according to the motor variable data to maintain the efficient operation of the motor. For example, when the motor load changes, the variable data such as the motor's stator current and speed will be affected. If the load suddenly increases, the motor speed will tend to decrease. At this time, the control system will detect these changing motor variable data, re-determine the synchronous rotation coordinate transformation information, and then adjust the first control strategy information and the second control strategy information. The torque output may be increased and the magnetic field strength may be adjusted to keep the motor speed stable near the set value to ensure that the motor can operate at a higher efficiency under different load conditions.

[0058] During the deceleration process of the motor, the control strategy will also change accordingly. For example, when the motor needs to stop, the control system will determine the synchronous rotation coordinate transformation information of the deceleration process based on the current motor variable data. If the motor is decelerated by energy-consuming braking, the first control strategy information may be adjusted to reduce the torque output and reasonably adjust the magnetic field component. For example, the torque output is reduced by gradually reducing the quadrature-axis current component, and the direct-axis current component is adjusted to control the magnetic field strength according to the motor speed and residual energy to ensure that the motor can decelerate smoothly and avoid abnormal situations such as sudden stops or excessive reversals.

[0059] In summary, the present invention provides a comprehensive and effective solution for the control of embedded permanent magnet synchronous motors. Through the close cooperation of various links, the performance potential of the motor is fully tapped, the operating efficiency and reliability of the motor are improved, and it can adapt to the needs of the motor under different operating conditions. Whether it is starting, accelerating, steady-state operation or deceleration, the control strategy can be adjusted in time according to the changes in the motor variable data to ensure the optimization of the motor performance.

[0060] The implementation process of the present invention is a continuous and dynamic process. During the entire life cycle of the motor, the control system continuously repeats the steps S110-S130, and continuously optimizes the control strategy according to the real-time changes of the motor variable data, thereby ensuring that the embedded permanent magnet synchronous motor is always in the best operating state. This control method not only improves the operating performance of the motor, but also prolongs the service life of the motor to a certain extent, reduces the operating cost of the motor, and has high practical value and broad application prospects.

[0061] In a preferred embodiment, determining the synchronous rotation coordinate transformation information from the motor variable data includes: generating derived motor variable data through the motor variable data, wherein the derived motor variable data includes the synchronous rotation coordinate transformation information, magnetic field component information, and timing control characteristics of the synchronous rotation coordinate transformation information and the magnetic field component information relative to the embedded permanent magnet synchronous motor to be controlled, and the magnetic field component information is used to optimize the synchronous rotation coordinate transformation information.

[0062] Further, generating derived motor variable data through the motor variable data includes: preprocessing the motor variable data using a decision control network to determine the synchronous rotation coordinate transformation information, the magnetic field component information and the derived motor variable data from the motor variable data.

[0063] Furthermore, generating the first control strategy information and component characteristics of the synchronous rotation coordinate transformation information based on the motor variable data includes: using a long short-term memory network to generate the first control strategy information based on the motor variable data; based on the generated first control strategy information, using the long short-term memory network to determine the weight semantics of the first control strategy information, the weight semantics indicating the reference confidence of the synchronous rotation coordinate transformation information in the first control strategy information; adjusting the weight semantics to generate the component characteristics of the synchronous rotation coordinate transformation information.

[0064] In this preferred embodiment, the entire technical solution is a highly integrated and complex system, with each link being closely related, and is intended to achieve more precise and effective control of the controlled embedded permanent magnet synchronous motor.

[0065] First, we focus on the key operation of determining synchronous rotation coordinate transformation information from motor variable data. Generating derived motor variable data from motor variable data is a multi-dimensional and in-depth process. Motor variable data covers a wealth of information, such as the values ​​of parameters such as stator current, stator voltage, motor speed, and their changing trends over time. For example, the stator current presents a series of values ​​at different times, such as 5 amperes at t1, 7 amperes at t2, and 8 amperes at t3, etc. These values ​​are all organic components of motor variable data. At the same time, the stator voltage may remain at around 380 volts in a certain stable operation stage, but it may also fluctuate slightly with changes in motor load or other factors, such as a short drop to 370 volts when the load suddenly increases. The same is true for motor speed, which may gradually increase from 1000 rpm to 1200 rpm in a certain operating range.

[0066] In the process of generating derived motor variable data from these motor variable data, the decision control network plays a core preprocessing role. The decision control network has a complex algorithm structure and can perform in-depth mining and analysis on motor variable data. For example, for the processing of stator current values, it does not simply record these values, but analyzes the motor operating status information behind the current value changes based on its internal algorithm logic. When the stator current changes from 5 amperes to 7 amperes, the decision control network can determine that this may be due to an increase in motor load or other related factors, and preprocess the motor variable data based on this judgment.

[0067] In the derived motor variable data, synchronous rotating coordinate transformation information, magnetic field component information, and the timing control characteristics based on the two relative to the embedded permanent magnet synchronous motor to be controlled are all of great significance. Synchronous rotating coordinate transformation information is a key element to realize vector control. It converts the motor variables in the three-phase stationary coordinate system into variables in the synchronous rotating coordinate system, thereby simplifying the mathematical model of the motor for easy control. The magnetic field component information provides support for optimizing the synchronous rotating coordinate transformation information from the perspective of the magnetic field. For example, the magnetic field component information may be reflected in the numerical form of magnetic field intensity. For example, under a certain operating state, the magnetic field intensity is 0.3 Tesla, which will affect some parameter adjustments in the synchronous rotating coordinate transformation information. The timing control characteristics take into account the time sequence factors in the operation of the motor, because the various variables of the motor have different relationships and influences at different time points. For example, at the beginning of the motor startup, the process of establishing the magnetic field has a specific timing relationship with the rise of the stator current, and this timing relationship is crucial to the control of the entire motor.

[0068] We further delve into the process of generating the first control strategy information and component characteristics of the synchronous rotation coordinate transformation information based on the motor variable data. The long short-term memory network plays an irreplaceable role in this process. The advantage of the long short-term memory network is that it can handle the timing information in the motor variable data very well, which is very critical for motor control. For example, considering the operation of the motor over a long period of time, the stator current may show complex periodic changes, which may be affected by the periodic changes of the motor load, the periodic influence of the power supply voltage fluctuation and other factors. The long short-term memory network can capture the change law on this long time scale, thereby providing a strong basis for generating the first control strategy information.

[0069] For example, during a long operation cycle of the motor, the stator current fluctuates with a cycle of 10 minutes, ranging from 5 amps to 10 amps. Based on this long-term current change pattern, combined with other motor variable data, such as the corresponding changes in motor speed in the same cycle (perhaps a periodic change from 800 rpm to 1200 rpm), the long short-term memory network generates the first control strategy information, which may take different control measures for current and speed changes at different stages. For example, in the current rising stage, a strategy of appropriately increasing the magnetic field strength to maintain a stable torque output may be adopted; in the current falling stage, the torque current component may be adjusted to improve motor efficiency, etc.

[0070] Based on the generated first control strategy information, the long short-term memory network determines the weight semantics of the first control strategy information, and this weight semantics indicates the reference confidence of the synchronous rotation coordinate transformation information in the first control strategy information. This concept is relatively abstract but very important. For example, in a certain specific operating state, the weight semantics may be quantified into a numerical value, such as 0.6. This value indicates that in the current first control strategy information, the synchronous rotation coordinate transformation information has a reference confidence of 60%. The determination of this value is the result of the long short-term memory network comprehensively considering many factors, including the current state of the motor variable data, the historical operating data of the motor, the current first control strategy information, and the physical characteristics of the motor itself.

[0071] Adjusting the weight semantics to generate the component features of the synchronous rotating coordinate transformation information is a process of refined control. When the weight semantics is adjusted from 0.6 to 0.5, it means that the importance of the synchronous rotating coordinate transformation information in the entire control strategy is relatively reduced. This adjustment will directly affect the change of component characteristics. For example, originally in the first control strategy information, the distribution ratio of the torque current component and the magnetic field current component corresponding to the synchronous rotating coordinate transformation information may be 3:2. After the weight semantics are adjusted, this distribution ratio may become 2:3, thereby changing the control method of the motor. This adjustment is to make the control strategy more adapted to the actual operation needs of the motor and to improve the flexibility and accuracy of motor control.

[0072] In the actual operation of the motor, these different operating links are intertwined and work together. This is a very typical scenario when the motor faces a sudden change in load. For example, the motor was originally in a stable operating state, and the load suddenly increased by 50%. This sudden change in load will immediately cause changes in the motor variable data, such as the stator current will increase rapidly. The stator current, which was originally 8 amps at time t1, may quickly rise to 12 amps at time t2 after the load increases. At the same time, the motor speed may have a downward trend, and the speed that was originally stable at 1000 rpm may drop to 900 rpm.

[0073] In this case, the decision control network quickly regenerates the derived motor variable data based on the new motor variable data. The decision control network is able to identify the two key changes of the sudden increase in stator current and the decrease in motor speed, and take these new conditions into account when generating the derived motor variable data. For example, the new magnetic field component information may be adjusted according to the increase in load, such as increasing the magnetic field strength from the original 0.3 Tesla to 0.4 Tesla to provide a stronger magnetic field to cope with the increased load.

[0074] At the same time, the long short-term memory network will also regenerate the first control strategy information based on the new motor variable data and the updated derived motor variable data. Due to the increase in load, the long short-term memory network may adjust the control strategy to pay more attention to the torque output to overcome the increased load. For example, in the new first control strategy information, the proportion of the torque current component may be increased from the original 30% to 40%. At the same time, the long short-term memory network will redefine the weight semantics and adjust the component characteristics. The weight semantics may be adjusted from the original 0.6 to 0.7, indicating that the reference confidence of the synchronous rotation coordinate transformation information in the new control strategy has been improved. Accordingly, the distribution ratio of the torque current component and the magnetic field current component corresponding to the synchronous rotation coordinate transformation information may be adjusted again, for example, to 4:1, to better adapt to the motor operation requirements after the load increases.

[0075] This collaborative working mechanism plays a key role in different operating stages of the motor. In the startup phase, the motor variable data has unique characteristics. The stator current gradually increases from the initial 0 amperes, and the motor speed increases from 0 rpm. At this time, the derived motor variable data generated by the decision control network based on these initial motor variable data provides an important control basis for the startup process. For example, the magnetic field component information may be set to a suitable initial value, such as 0.2 Tesla, to help establish the initial magnetic field and promote the smooth start of the motor. The first control strategy information generated by the long short-term memory network will focus on making the motor start smoothly, and may adopt a strategy of slowly increasing the torque current component. The weight semantics and component features will also be adjusted accordingly. For example, the weight semantics may be set to 0.8 to ensure that the synchronous rotation coordinate transformation information has a high reference confidence in the startup phase, while the component features may set the ratio of the torque current component to the magnetic field current component to 1:1 to balance the needs of magnetic field establishment and torque output.

[0076] During the acceleration phase, the motor's operating state changes rapidly. As the motor speed continues to increase, the motor variable data also changes rapidly. The stator current may continue to increase, and the motor speed may rise rapidly from 1000 rpm to 1500 rpm. The decision control network and the long short-term memory network need to continuously update the processing results to adapt to this rapid change. The decision control network will regenerate the derived motor variable data based on the new motor variable data. For example, the magnetic field component information may be dynamically adjusted as the speed increases to ensure that the magnetic field strength of the motor during the acceleration process can meet the torque requirements. The first control strategy information generated by the long short-term memory network will be adjusted according to the acceleration requirements, and may further increase the proportion of the torque current component to provide greater torque, so that the motor can accelerate quickly and stably. At the same time, the weight semantics and component features will also be adjusted accordingly. The weight semantics may fluctuate between 0.6 and 0.8 at different stages of the acceleration process, and the proportion of the torque current component and the magnetic field current component in the component features will also change continuously to adapt to various requirements during the acceleration process.

[0077] During the steady-state operation phase, although the motor variable data is relatively stable, there are still some small fluctuations. For example, the load may fluctuate slightly by about 10%, which will cause small changes in the stator current and motor speed. The decision control network and the long short-term memory network will continuously monitor the motor variable data and make fine adjustments to these small fluctuations. The decision control network will slightly adjust the derived motor variable data according to the changes in the motor variable data caused by load fluctuations. For example, the magnetic field component information may be adjusted in a small range of about 0.3 Tesla. The first control strategy information generated by the long short-term memory network will also be fine-tuned, such as adjusting the ratio of the torque current component and the magnetic field current component to maintain the steady-state operation of the motor. The weight semantics and component features will also be fine-tuned accordingly to ensure that the motor can still operate stably and maintain high operating efficiency under load fluctuations.

[0078] During the deceleration phase, the motor variable data presents a specific trend of change. The stator current will gradually decrease, and the motor speed will gradually decrease from a higher value. For example, the motor speed will gradually decrease from 1500 rpm, and the stator current will gradually decrease from 10 amps. The decision control network will generate derivative motor variable data suitable for deceleration based on these changing motor variable data. For example, the magnetic field component information may be adjusted as the speed decreases to ensure that the magnetic field of the motor can smoothly transition during the deceleration process. The first control strategy information generated by the long short-term memory network will focus on making the motor decelerate smoothly, and may gradually reduce the proportion of the torque current component. The weight semantics and component features will also be adjusted accordingly. For example, the weight semantics may gradually decrease from 0.6 to 0.5, and the proportion of the torque current component and the magnetic field current component in the component features will also change accordingly to achieve smooth deceleration of the motor.

[0079] In this way, an efficient and accurate motor control system is constructed by introducing a decision control network and a long short-term memory network. When the decision control network processes the motor variable data to generate the derived motor variable data, the magnetic field component information can optimize the synchronous rotation coordinate transformation information and improve the basic accuracy of the vector control. The long short-term memory network processes the timing information in the motor variable data, providing a basis for generating the first control strategy information that is more in line with the motor operation law. Determining the weight semantics and adjusting the component characteristics further improves the refinement of the control. This collaborative working mechanism enables the motor to quickly and accurately adjust the control strategy when facing various situations such as load changes and conversions between different operating stages, effectively improving the accuracy, stability and adaptability of the motor control, thereby improving the overall performance and operating efficiency of the motor, reducing energy waste and extending the service life of the motor.

[0080] In an alternative embodiment, the method further comprises: determining discrete state prediction information based on the second control strategy information, wherein the discrete state prediction information indicates a discrete state trend of the torque control decision parameter compared to the second control strategy information.

[0081] In the next step, the control processing of the embedded permanent magnet synchronous motor to be controlled is performed according to the first control strategy information and the component characteristics, and the second control strategy information, including: extracting the torque control decision parameters from the second control strategy information according to the discrete state prediction information; generating a control execution task label according to the component characteristics, the first control strategy information and the torque control decision parameters to control the embedded permanent magnet synchronous motor to be controlled.

[0082] Correspondingly, a control execution task label is generated based on the component characteristics, the first control strategy information and the torque control decision parameters to control the embedded permanent magnet synchronous motor to be controlled, including: based on three dimensional descriptions corresponding to the control dimension characteristics of the first control strategy information, the control dimension characteristics of the second control strategy information, and the linkage dimension characteristics between the second control strategy information and the first control strategy information, the embedded permanent magnet synchronous motor to be controlled is controlled through multiple task node labels.

[0083] Furthermore, controlling the embedded permanent magnet synchronous motor to be controlled through multiple task node tags includes: in each parameter control cycle corresponding to each task node tag, according to the dimension description corresponding to the task node tag in the three dimensional descriptions, performing vector control based on coordinate transformation on the embedded permanent magnet synchronous motor to be controlled; wherein, for the first task node tag among the multiple task node tags, the dimension description corresponding to the first task node tag is the three dimensional description; for the second task node tag among the multiple task node tags, the dimension description corresponding to the second task node tag is the dimension description corresponding to the control dimensional characteristics of the first control strategy information, and the dimension description corresponding to the control dimensional characteristics of the second control strategy information; for the third task node tag among the multiple task node tags, the dimension description corresponding to the third task node tag is the dimension description corresponding to the control dimensional characteristics of the second control strategy information.

[0084] In this alternative embodiment, the entire technical solution revolves around a series of complex and interrelated operations for controlling the motor.

[0085] First, the discrete state prediction information is determined based on the second control strategy information. The second control strategy information contains many key elements related to motor control, among which the torque control decision parameter is an important part. The discrete state prediction information indicates the discrete state trend of the torque control decision parameter compared with the second control strategy information. For example, in some cases, the torque control decision parameter may show a discrete change trend as the motor operating state changes. For example, within a specific operating period, the torque control decision parameter gradually changes from one numerical range to another numerical range. This numerical jump or stage-by-stage change reflects the discrete state trend. This discrete state trend may be affected by many factors, such as sudden changes in motor load, fluctuations in the magnetic field inside the motor, etc.

[0086] Next, based on the discrete state prediction information, the torque control decision parameter is extracted from the second control strategy information. This operation is a precise extraction based on the discrete state prediction information determined previously. Since the discrete state prediction information reflects the changing trend of the torque control decision parameter in the second control strategy information, the corresponding torque control decision parameter can be accurately found from the second control strategy information based on this trend. For example, if the discrete state prediction information indicates that the torque control decision parameter is on an upward trend at a certain moment and is about to reach a specific numerical range, then the torque control decision parameter that is on an upward trend and meets the numerical range requirements can be accurately extracted from the second control strategy information.

[0087] Then, a control execution task label is generated based on the component characteristics, the first control strategy information and the torque control decision parameter to control the embedded permanent magnet synchronous motor to be controlled. The component characteristics here are important characteristics related to the synchronous rotation coordinate transformation information that have been determined before, and the first control strategy information is also generated based on multiple factors such as motor variable data. For example, the component characteristics may be reflected in the weight relationship between different control components. For example, in a certain operating state, the torque component weight is 0.6 and the magnetic field component weight is 0.4. The first control strategy information may determine different control directions and focuses based on the current operating state of the motor, such as speed, current, etc. The torque control decision parameter is the numerical information specifically related to torque control. By combining this information, a control execution task label is generated. This label is like an instruction set that guides the control operation of the motor.

[0088] Furthermore, based on the three dimensional descriptions corresponding to the control dimension characteristics of the first control strategy information, the control dimension characteristics of the second control strategy information, and the linkage dimension characteristics between the second control strategy information and the first control strategy information, the embedded permanent magnet synchronous motor to be controlled is controlled through multiple task node labels. These three dimensional descriptions cover a wealth of information. The control dimension characteristics of the first control strategy information may involve different aspects of motor control, such as torque control, magnetic field control, etc., which are specifically reflected in the first control strategy information. For example, from the perspective of the control dimension characteristics, the first control strategy information may specify the control method of torque and magnetic field in different speed ranges. For example, in the low speed range, more emphasis is placed on the stable establishment of the magnetic field, while in the high speed range, more emphasis is placed on the efficient output of torque.

[0089] The control dimension characteristics of the second control strategy information also contain important content related to the torque control decision parameters. For example, it may specify the adjustment method of the torque control decision parameters under different load conditions. Under light load, the torque control decision parameters may change according to one rule, and under heavy load, according to another rule. The linkage dimension characteristics between the second control strategy information and the first control strategy information reflect the relationship and collaborative working mode between the two. For example, when the first control strategy information emphasizes magnetic field control under a certain operating state, how does the second control strategy information cooperate with it to adjust the torque control decision parameters?

[0090] Control processing is performed through multiple task node tags, and different operation methods are used for different task node tags. For the first task node tag among multiple task node tags, the dimension description corresponding to the first task node tag is a three-dimensional description. This means that under this task node tag, all dimensional description information will be comprehensively considered to perform control operations. For example, in the operation corresponding to this task node tag, the torque and magnetic field control methods in the control dimensional characteristics of the first control strategy information, the adjustment methods of the torque control decision parameters in the control dimensional characteristics of the second control strategy information under different circumstances, and the collaborative working methods in the linkage dimensional characteristics between the two will be considered at the same time. In the specific control process, precise control operations may be performed according to the requirements in the three-dimensional description based on the current operating state of the motor, such as a speed of 1000 revolutions per minute, a load of 60% of the rated load, etc.

[0091] For the second task node label among the multiple task node labels, the dimension description corresponding to the second task node label is the dimension description corresponding to the control dimension feature of the first control strategy information, and the dimension description corresponding to the control dimension feature of the second control strategy information. Under this task node label, the focus is on the control dimension features of the first control strategy information and the second control strategy information. For example, in the operation corresponding to this task node label, when the running state of the motor changes, such as the speed increases from 1000 rpm to 1200 rpm, the magnetic field control mode will be adjusted according to the control dimension feature of the first control strategy information, and the torque control decision parameters will be adjusted according to the control dimension feature of the second control strategy information, while the linkage dimension feature between the two is less considered.

[0092] For the third task node label among the multiple task node labels, the dimension description corresponding to the third task node label is the dimension description corresponding to the control dimension feature of the second control strategy information. Under this task node label, the focus is on the control dimension feature of the second control strategy information. For example, when the motor is in a certain operating stage, such as when running close to the rated speed, the control operation is mainly performed according to the torque control decision parameter adjustment method in the control dimension feature of the second control strategy information, without considering the control dimension feature of the first control strategy information and the linkage dimension feature between the two.

[0093] In each parameter control cycle corresponding to each task node label, according to the dimension description corresponding to the task node label in the three-dimensional description, the vector control based on coordinate transformation is performed on the controlled embedded permanent magnet synchronous motor. In the parameter control cycle, vector control operations are performed according to different task node labels and corresponding dimension descriptions. For example, in a parameter control cycle, if it is the first task node label, vector control is performed according to the comprehensive three-dimensional description, including operations such as adjusting the direct-axis and quadrature-axis components of the stator current. For example, in this process, according to the current operating state of the motor and the requirements of the three-dimensional description, the direct-axis current component is set to 3 amperes and the quadrature-axis current component is set to 4 amperes to achieve appropriate torque output and magnetic field strength. If it is the second task node label, vector control is performed according to the dimension description corresponding to the control dimension feature of the first control strategy information and the control dimension feature of the second control strategy information. If it is the third task node label, vector control is performed according to the dimension description corresponding to the control dimension feature of the second control strategy information.

[0094] In this way, the torque control decision parameters are accurately extracted by determining the discrete state prediction information, and then the control execution task label is generated using the component characteristics, the first control strategy information and the torque control decision parameters. Motor control is performed through three-dimensional descriptions and multiple task node labels. This method can control the controlled embedded permanent magnet synchronous motor more carefully and accurately. Vector control is performed in each parameter control cycle according to different task node labels and corresponding dimensional descriptions, which can adapt to the needs of the motor under different operating conditions, improve the flexibility and accuracy of control, optimize the operating performance of the motor, and ensure that the motor can operate efficiently and stably under various working conditions.

[0095] In an exemplary technical solution, the method further includes: in response to a control adjustment instruction for the embedded permanent magnet synchronous motor to be controlled, performing sensorless control processing on the embedded permanent magnet synchronous motor to be controlled.

[0096] Among them, the sensorless control processing includes: obtaining an electromotive force statistical data sequence, wherein the electromotive force statistical data sequence includes X consecutive groups of electromotive force statistical data, and X is an integer greater than or equal to 1; obtaining a loss statistical data sequence according to the electromotive force statistical data sequence, wherein the loss statistical data sequence includes X consecutive groups of loss statistical data; based on the electromotive force statistical data sequence, obtaining an electromotive force vector distribution sequence through a first vector mining branch included in an extended Kalman filter network, wherein the electromotive force vector distribution sequence includes X electromotive force vector distributions; based on the loss statistical data sequence, obtaining a loss vector distribution sequence through a second vector mining branch included in the extended Kalman filter network, wherein the loss vector distribution sequence includes X loss vector distributions; based on the electromotive force vector distribution sequence and the loss vector distribution sequence, obtaining a rule discrimination weight corresponding to the electromotive force statistical data through a control output branch included in the extended Kalman filter network; and determining the sensorless control mechanism of the embedded permanent magnet synchronous motor to be controlled according to the rule discrimination weight.

[0097] In this exemplary technical solution, the entire technical process revolves around the sensorless control processing of the controlled embedded permanent magnet synchronous motor, which is a highly complex process with tightly coupled parts.

[0098] When a control adjustment instruction for the embedded permanent magnet synchronous motor to be controlled is received, the sensorless control processing mechanism starts to operate. First, it is crucial to obtain the sequence of electromotive force statistics. This sequence contains X consecutive sets of electromotive force statistics, where X is an integer greater than or equal to 1. The electromotive force statistics are a key reflection of the motor's operating status, and they contain rich electromagnetic information inside the motor.

[0099] Taking an actual motor operation scenario as an example, at a certain stage of stable operation of the motor, we may obtain a set of data in the sequence of electromotive force statistics. The electromotive force value in this set of data may be 5 volts, and this value does not exist in isolation. It is associated with multiple operating parameters of the motor. For example, it may be closely related to factors such as the motor speed, the number of turns of the stator winding, the magnetic field strength of the permanent magnet, and the magnetic circuit structure inside the motor. During the operation of the motor, due to the change of the magnetic field and the cutting movement of the conductor, an electromotive force will be generated, and this 5 volt value is generated under specific motor structure and operating conditions. As the motor runs, we will continue to obtain such electromotive force statistics to form a sequence containing multiple sets of data.

[0100] Obtaining a loss statistics sequence based on an electromotive force statistics sequence is an important part of this technical solution. The loss statistics sequence also contains X sets of continuous loss statistics. The loss of a motor is a complex phenomenon, which includes multiple types such as copper loss and iron loss. For example, in the process of converting from electromotive force statistics to loss statistics, when the electromotive force is 5 volts, according to the resistance characteristics of the motor, if the resistance of the stator winding is 1 ohm, the current at this voltage can be calculated through a specific electrical relationship, and then the copper loss part can be calculated based on the square of the current multiplied by the resistance. At the same time, iron loss is related to factors such as the core material of the motor and the alternating frequency of the magnetic field. Through some characteristics in the electromotive force statistics, such as frequency components, combined with knowledge such as the hysteresis characteristics of the core material, the iron loss can be estimated. In this way, a set of loss statistics can be obtained, and as the operating state of the motor changes, these data are continuously accumulated to form a loss statistics sequence.

[0101] Based on the sequence of electromotive force statistical data, the first vector mining branch included in the extended Kalman filter network is used to obtain the electromotive force vector distribution sequence, which contains X electromotive force vector distributions. The extended Kalman filter network is a powerful signal processing tool, and its first vector mining branch has unique functions. Taking the previously mentioned set of data with an electromotive force of 5 volts as an example, the first vector mining branch not only focuses on the value itself, but also mines the vector information hidden behind it. For example, in the three-phase coordinate system of the motor, this electromotive force may have different distributions between different phases. The electromotive force vector component of one phase may be 2 volts and that of another phase may be 3 volts. This distribution in different phases or different dimensions is extracted by the first vector mining branch to form an electromotive force vector distribution. As each set of electromotive force statistical data is processed, a sequence containing X electromotive force vector distributions is obtained.

[0102] Based on the loss statistical data sequence, the loss vector distribution sequence is obtained by the second vector mining branch included in the extended Kalman filter network, wherein the loss vector distribution sequence includes X loss vector distributions. The second vector mining branch processes the loss statistical data in a similar way to the first vector mining branch processes the electromotive force statistical data. For example, for different types of losses such as copper loss and iron loss, their distribution is different in different parts of the motor or under different operating conditions. For example, at a certain operating moment of the motor, the copper loss is mainly concentrated in a certain section of the stator winding, and its vector distribution may be manifested as a large loss intensity in this section and relatively small in other sections. The iron loss may be more obvious in a certain area of ​​the core. This loss distribution under the influence of space or different factors is mined by the second vector mining branch to form a loss vector distribution. With the processing of each set of loss statistical data, a sequence containing X loss vector distributions is constructed.

[0103] Based on the electromotive force vector distribution sequence and the loss vector distribution sequence, the rule discrimination weight corresponding to the electromotive force statistical data is obtained through the control output branch included in the extended Kalman filter network. This process is one of the key links of the entire technical solution. The control output branch of the extended Kalman filter network comprehensively considers various information of the electromotive force vector distribution and the loss vector distribution. For example, when the electromotive force vector distribution shows that the electromotive force vector component of a certain phase has a large fluctuation over a period of time, and the loss vector distribution shows that during this fluctuation period, the loss of a certain part of the motor also changes accordingly, the control output branch will determine the rule discrimination weight based on the pre-set algorithm and model, combining this information. This weight may be a value between 0 and 1, for example, this value is 0.6 in a certain operating state. This value reflects the importance of electromotive force statistical data in determining the sensorless control mechanism in the current motor operating state relative to other factors (such as loss, etc.).

[0104] Finally, the sensorless control mechanism of the embedded permanent magnet synchronous motor to be controlled is determined according to the rule discrimination weight. This sensorless control mechanism is the ultimate goal of the entire sensorless control process. When the rule discrimination weight is 0.6, it means that the electromotive force statistics occupy a relatively important position in the sensorless control mechanism. For example, in this case, the sensorless control mechanism may adjust the control strategy of the motor more based on the electromotive force vector distribution. If the electromotive force vector distribution shows that the electromotive force of a certain phase is low at a specific moment, the magnetic field strength or torque output of the phase may be changed by adjusting the distribution of the stator current to achieve the purpose of optimizing the operation of the motor. At the same time, the loss vector distribution will also affect the control mechanism to a certain extent. If the loss vector distribution indicates that the loss of a certain part is too high, the sensorless control mechanism may adjust the operating parameters of the motor, such as reducing the speed or adjusting the current frequency, to reduce the loss of that part.

[0105] In the entire sensorless control process, there is a close logical connection between each link. The electromotive force statistical data sequence is used as the starting point, which provides the basis for the acquisition of the subsequent loss statistical data sequence, electromotive force vector distribution sequence and loss vector distribution sequence. The electromotive force vector distribution sequence and the loss vector distribution sequence together provide the basis for determining the rule discrimination weight, and finally determine the sensorless control mechanism. This processing method based on the extended Kalman filter network can deeply explore the electromotive force and loss information inside the motor, and effectively integrate this information into the sensorless control mechanism.

[0106] This sensorless control mechanism plays an important role in different operation stages of the motor. In the motor startup stage, the electromotive force statistical data sequence may show a gradually rising trend. As the motor speed increases from 0, the electromotive force also gradually increases from a lower value. The loss statistical data sequence obtained based on the electromotive force statistical data sequence at this stage may show that the copper loss is relatively large because the current is large during the startup process. Through the processing of the extended Kalman filter network, the electromotive force vector distribution sequence and the loss vector distribution sequence obtained will reflect the special situation of the startup stage. For example, the electromotive force vector distribution may show the imbalance of the three-phase electromotive force at the moment of startup, and the loss vector distribution may indicate that the copper loss is mainly concentrated in the starting part of the stator winding. The rule discrimination weights determined based on this information may make the sensorless control mechanism pay more attention to adjusting the current distribution in the startup stage to balance the electromotive force and reduce the copper loss to ensure that the motor can start smoothly.

[0107] During the acceleration phase of the motor, the EMF statistical data sequence changes faster, and as the speed increases rapidly, the EMF value also increases rapidly. At this time, the loss statistical data sequence obtained based on the EMF statistical data sequence may show that the iron loss gradually increases because the alternating frequency of the magnetic field increases. The EMF vector distribution sequence and the loss vector distribution sequence obtained after processing by the extended Kalman filter network will reflect the dynamic changes during this acceleration process. For example, the EMF vector distribution may show that the difference in EMF between different phases gradually increases during the acceleration process, and the loss vector distribution may indicate that the iron loss increases rapidly in certain areas of the core. The rule discrimination weights determined based on this information may enable the sensorless control mechanism to adjust the control strategy of the magnetic field during the acceleration phase to balance the EMF and control the growth of iron loss, so that the motor can accelerate quickly and stably.

[0108] During the steady-state operation of the motor, the EMF statistical data sequence is relatively stable, but there are still some small fluctuations. These fluctuations may be caused by small changes in the load or small fluctuations in the power supply voltage. The loss statistical data sequence obtained based on the EMF statistical data sequence will also have corresponding small fluctuations, mainly small changes in copper loss and iron loss. The EMF vector distribution sequence and loss vector distribution sequence obtained after processing by the extended Kalman filter network will reflect these small fluctuations. For example, the EMF vector distribution may show that the EMF of a certain phase fluctuates up and down within a certain range, and the loss vector distribution may show that the copper loss varies slightly in different parts of the stator winding. The rule discrimination weights determined based on this information may enable the sensorless control mechanism to fine-tune the operating parameters of the motor during the steady-state operation stage, such as slightly adjusting the current size or frequency, to maintain stable operation of the motor and keep low losses.

[0109] During the deceleration phase of the motor, the EMF statistical data sequence shows a downward trend, and the EMF value gradually decreases as the motor speed decreases. The loss statistical data sequence obtained based on the EMF statistical data sequence may show that copper loss becomes the main loss component again because the current changes during the deceleration process. The EMF vector distribution sequence and the loss vector distribution sequence obtained after processing by the extended Kalman filter network will reflect the characteristics of the deceleration phase. For example, the EMF vector distribution may show that the imbalance of the three-phase EMF reappears during the deceleration process, and the loss vector distribution may indicate that the copper loss is mainly concentrated in certain specific parts of the stator winding. The rule discrimination weights determined based on this information may enable the sensorless control mechanism to adjust the current recovery strategy or the magnetic field weakening method during the deceleration phase to smoothly reduce the motor speed and reduce losses.

[0110] In this way, by extending the Kalman filter network to perform multi-branch processing on the electromotive force statistical data sequence and the loss statistical data sequence, the electromotive force vector distribution sequence, the loss vector distribution sequence and the rule discrimination weight are deeply mined, and then the sensorless control mechanism is determined. This process comprehensively and meticulously utilizes the electromotive force and loss-related information of the motor, so that the sensorless control mechanism can accurately adapt to the needs of the motor in different operating stages. Without relying on sensors, effective control is carried out according to the complex electromagnetic state and energy loss conditions inside the motor, which greatly improves the accuracy and reliability of control, optimizes the operating efficiency of the motor, reduces the cost caused by sensor failure or the use of sensors, enhances the stability and adaptability of the motor control system, and provides an efficient and comprehensive solution for sensorless control of embedded permanent magnet synchronous motors.

[0111] The embodiment of the present invention can accurately construct the motor digital-analog data by obtaining the motor variable data to determine the synchronous rotation coordinate transformation information, laying a solid foundation for subsequent control. The first control strategy information and component characteristics and the second control strategy information are generated based on the motor variable data, and the control component weights can be carefully weighed to make the torque control decision parameters more in line with the actual needs of the motor. Finally, control processing is performed based on this information to achieve accurate and efficient control of the embedded permanent magnet synchronous motor. This control method helps to improve the operating efficiency of the motor, optimize torque output, enhance the adaptability of the motor to different working conditions, reduce energy consumption and extend the service life of the motor, and improve the overall performance of the motor.

[0112] An embodiment of the present invention provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the control method of the embedded permanent magnet synchronous motor is implemented.

[0113] An embodiment of the present invention provides a processor, which is used to run a program, wherein the control method of the embedded permanent magnet synchronous motor is executed when the program is run.

[0114] In the embodiment of the present invention, Figure 2 As shown, the embedded permanent magnet synchronous motor control system 100 includes at least one processor 101, and at least one memory 102 and a bus 103 connected to the processor 101; wherein the processor 101 and the memory 102 communicate with each other through the bus 103; the processor 101 is used to call the program instructions in the memory 102 to execute the above-mentioned embedded permanent magnet synchronous motor control method.

[0115] The present invention is described with reference to the flowchart and / or block diagram of the method, embedded permanent magnet synchronous motor control system (system), and computer program product according to the embodiment of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0116] In a typical configuration, the embedded permanent magnet synchronous motor control system includes one or more processors (CPU), memory and bus. The embedded permanent magnet synchronous motor control system may also include input / output interface, network interface and the like.

[0117] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip. The memory is an example of a computer-readable medium.

[0118] Computer readable media include permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage computer readable storage media or any other non-transmission media that can be used to store information that can be accessed by the embedded permanent magnet synchronous motor control system. As defined in this article, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carriers.

[0119] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or computer-readable storage medium that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements inherent to such process, method, commodity or computer-readable storage medium. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the presence of other identical elements in the process, method, commodity or computer-readable storage medium that includes the elements.

[0120] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0121] The above are only embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A control method for an embedded permanent magnet synchronous motor, characterized in that: The method is applied to an embedded permanent magnet synchronous motor control system, and the method comprises: Acquire motor variable data for making vector control decisions, and determine synchronous rotating coordinate transformation information from the motor variable data, wherein the synchronous rotating coordinate transformation information represents motor digital-analog data corresponding to the embedded permanent magnet synchronous motor to be controlled; wherein the synchronous rotating coordinate transformation information is obtained by performing synchronous rotating coordinate transformation on current variables and voltage variables of the embedded permanent magnet synchronous motor to be controlled in a three-phase stationary coordinate system, and in the synchronous rotating coordinate transformation process, the current variables and voltage variables are converted into a synchronous rotating coordinate system oriented by the motor rotor magnetic field; Generate first control strategy information and component characteristics of the synchronous rotating coordinate transformation information according to the motor variable data, and generate second control strategy information according to the synchronous rotating coordinate transformation information, wherein the component characteristics indicate the control component weight of the synchronous rotating coordinate transformation information relative to the first control strategy information, and the second control strategy information includes a torque control decision parameter corresponding to the synchronous rotating coordinate transformation information; wherein, when the first control strategy information indicates that the importance of torque control is higher than that of magnetic field control, the component characteristics indicate that the torque component weight in the synchronous rotating coordinate transformation information is higher than the magnetic field component weight; According to the first control strategy information and the component characteristics, and the second control strategy information, control processing is performed on the embedded permanent magnet synchronous motor to be controlled; The method further includes: in response to a control adjustment instruction for the embedded permanent magnet synchronous motor to be controlled, performing a sensorless control process on the embedded permanent magnet synchronous motor to be controlled; The sensorless control processing includes: obtaining an electromotive force statistical data sequence of the embedded permanent magnet synchronous motor to be controlled, wherein the electromotive force statistical data sequence includes X consecutive groups of electromotive force statistical data, and X is an integer greater than or equal to 1; obtaining a loss statistical data sequence according to the electromotive force statistical data sequence, wherein the loss statistical data sequence includes X consecutive groups of loss statistical data; based on the electromotive force statistical data sequence, obtaining an electromotive force vector distribution sequence through a first vector mining branch included in an extended Kalman filter network, wherein the electromotive force vector distribution sequence includes X electromotive force vector distributions; based on the loss statistical data sequence Sequence, obtain a loss vector distribution sequence through the second vector mining branch included in the extended Kalman filter network, wherein the loss vector distribution sequence includes X loss vector distributions; based on the electromotive force vector distribution sequence and the loss vector distribution sequence, obtain the rule discrimination weight corresponding to the electromotive force statistical data through the control output branch included in the extended Kalman filter network; determine the sensorless control mechanism of the embedded permanent magnet synchronous motor to be controlled according to the rule discrimination weight; wherein the rule discrimination weight reflects the importance of the electromotive force statistical data relative to the loss in determining the sensorless control mechanism under the current motor operating state.

2. The method according to claim 1, characterized in that Determining synchronous rotation coordinate transformation information from the motor variable data includes: Generate derived motor variable data through the motor variable data, wherein the derived motor variable data includes the synchronous rotating coordinate transformation information, the magnetic field component information, and the timing control characteristics for the embedded permanent magnet synchronous motor to be controlled determined according to the synchronous rotating coordinate transformation information and the magnetic field component information, and the magnetic field component information is used to optimize the synchronous rotating coordinate transformation information; Among them, the timing control characteristics are used to characterize the time sequence factors in the motor operation process and guide the motor control from the perspective of timing relationship.

3. The method according to claim 2, characterized in that Generating derived motor variable data from the motor variable data includes: preprocessing the motor variable data using a decision control network to determine the derived motor variable data from the motor variable data.

4. The method according to claim 2, characterized in that The first control strategy information and component characteristics of generating the synchronous rotation coordinate transformation information according to the motor variable data include: Using a long short-term memory network, generating the first control strategy information according to the motor variable data; Determining weight semantics of the first control strategy information using the long short-term memory network according to the generated first control strategy information, wherein the weight semantics indicates a reference confidence of the synchronous rotation coordinate transformation information in the first control strategy information; The weight semantics are adjusted to generate component features of the synchronous rotation coordinate transformation information.

5. The method according to claim 1, characterized in that The method further includes determining discrete state prediction information based on the second control strategy information, wherein the discrete state prediction information indicates a discrete state trend of the torque control decision parameter compared to the second control strategy information.

6. The method according to claim 5, characterized in that The controlling process of the embedded permanent magnet synchronous motor to be controlled according to the first control strategy information, the component characteristics, and the second control strategy information includes: extracting the torque control decision parameter from the second control strategy information according to the discrete state prediction information; A control execution task label is generated according to the component characteristics, the first control strategy information and the torque control decision parameter to perform control processing on the embedded permanent magnet synchronous motor to be controlled.

7. The method according to claim 6, characterized in that Generating a control execution task label according to the component characteristics, the first control strategy information and the torque control decision parameter to control the embedded permanent magnet synchronous motor to be controlled includes: controlling the embedded permanent magnet synchronous motor to be controlled through a plurality of task node labels according to three dimensional descriptions corresponding to the control dimensional characteristics of the first control strategy information, the control dimensional characteristics of the second control strategy information, and the linkage dimensional characteristics between the second control strategy information and the first control strategy information; Controlling the embedded permanent magnet synchronous motor to be controlled by multiple task node tags includes: in each parameter control cycle corresponding to each task node tag, according to the dimension description corresponding to the task node tag in the three dimensional descriptions, performing vector control based on coordinate transformation on the embedded permanent magnet synchronous motor to be controlled; for a first task node tag among the multiple task node tags, the dimension description corresponding to the first task node tag is the three dimensional description; for a second task node tag among the multiple task node tags, the dimension description corresponding to the second task node tag is the dimension description corresponding to the control dimensional feature of the first control strategy information, and the dimension description corresponding to the control dimensional feature of the second control strategy information; for a third task node tag among the multiple task node tags, the dimension description corresponding to the third task node tag is the dimension description corresponding to the control dimensional feature of the second control strategy information; Among them, the control dimension characteristics of the first control strategy information include the implementation methods of torque control and magnetic field control in the first control strategy information; the control dimension characteristics of the second control strategy information include the adjustment methods of torque control decision parameters under different load conditions; the linkage dimension characteristics reflect the relationship and collaborative working method between the first control strategy information and the second control strategy information.

8. An embedded permanent magnet synchronous motor control system, characterized in that: It comprises a processor and a memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call the program instructions in the memory to execute the control method of the embedded permanent magnet synchronous motor described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that: A program is stored thereon, and when the program is executed by a processor, the control method of the embedded permanent magnet synchronous motor described in any one of claims 1-7 is implemented.

Images