Switching frequency switching method and system based on Markov chain, and vehicle

Abstract

The invention discloses a switching frequency switching method and system based on a Markov chain and a vehicle, and relates to the technical field of vehicle control. The method comprises the steps that after random frequency conversion is started, working condition parameters are obtained, and the Markov chain state matched with the working condition parameters is determined based on a predefined Markov chain and a working condition-state mapping rule; wherein the Markov chain comprises a plurality of Markov chain states, and each Markov chain state is provided with an independent switching frequency interval and a working condition parameter boundary; generating a random switching frequency in the determined switching frequency interval of the Markov chain state; and updating the working condition parameters, re-determining the Markov chain state, obtaining the state transition probability according to the Markov chain state before and after re-determination and a pre-constructed state transition probability matrix, and regenerating the random switching frequency if the state is transited. According to the invention, dynamic and adaptive adjustment of the switching frequency of the vehicle can be realized.

Description

Technical Field

[0001] The present invention relates to vehicle control, and in particular to a switching frequency switching method and system based on a Markov chain, and a vehicle. Background Art

[0002] The continuous development of the electric vehicle industry has also driven the rapid expansion of the motor industry. Since its introduction, space vector pulse width modulation (SVPWM) technology has been widely used in the motor control field due to its technical advantages such as excellent harmonic characteristics and high voltage utilization. SVPWM technology modulates the desired waveform by frequently controlling the power module to turn on and off. However, long-term engineering practice has shown that SVPWM technology produces harmonic clusters at the switching frequency and its multiples. Specifically, the controller generates torque pulsations and a relatively harsh whine during motor operation. This whine is a high-frequency, high-order noise that affects the NVH performance of the entire vehicle. This has led to the discovery of a new PWM modulation strategy: the random switching frequency PWM modulation strategy.

[0003] The random switching frequency PWM modulation strategy changes the switching period and only the corresponding control signal, without changing the original topology. It can disperse the harmonic energy originally clustered at the switching frequency and its multiples to the surrounding frequency bands, reducing the harmonic energy at the switching frequency and its multiples. This not only alleviates the torque pulsation and noise problems caused by harmonics, but also significantly improves the electromagnetic compatibility performance of the motor system.

[0004] A typical strategy in existing technology solutions is to establish an NVH random frequency model corresponding to the NVH noise prominence range based on the NVH spectrum at the IGBT base frequency, thereby implementing random frequency control. In random frequency control, the IGBT switching frequency is controlled to randomly vary symmetrically around the base frequency f, so that the random frequency values ​​are normally distributed around the base frequency f, with the base frequency f at the center. Although this solution can adjust the control frequency, the normal distribution characteristics cause most random numbers to be close to the base frequency f, resulting in a weak random effect and the inability to achieve adaptive switching of the switching frequency based on the operating conditions. Summary of the Invention

[0005] In view of this, an object of the present invention is to provide a switching frequency switching method, system and vehicle based on a Markov chain, which can achieve dynamic and adaptive adjustment of the switching frequency of the vehicle.

[0006] A switching frequency switching method based on a Markov chain in the present invention includes: After random frequency conversion is enabled, operating condition parameters are obtained, and based on a predefined Markov chain and operating condition-state mapping rule, a Markov chain state that matches the operating condition parameters is determined; wherein the Markov chain includes a plurality of Markov chain states, each of which has an independent switching frequency interval and operating condition parameter boundary; the operating condition-state mapping rule is that if all operating condition parameters fall within a range defined by the operating condition parameter boundary of a certain Markov chain state, then it is determined that the current operating condition corresponds to the Markov chain state; Generate a random switching frequency within the switching frequency interval of the determined Markov chain state; Update the operating parameters and redetermine the Markov chain state, and obtain the state transition probability based on the Markov chain state before and after redetermination and the pre-built state transition probability matrix; The state transition is determined based on the comparison between the state transition probability and the preset threshold. If the state transition occurs, the random switching frequency is regenerated based on the Markov chain state after the transition.

[0007] Furthermore, the method further includes: constructing a state transition probability matrix, wherein constructing the state transition probability matrix specifically includes: Preset initial state transition probability matrix; Set the probability matrix update cycle, and based on the vehicle operation data, count the actual number of transitions and the total number of transitions for each Markov chain state in the previous probability matrix update cycle, and re-determine the state transition probability of each Markov chain state; Based on the re-determined state transition probabilities, the state transition probability matrix is ​​updated.

[0008] Furthermore, the random switching frequency is generated by using a random frequency conversion generation algorithm, and the random frequency conversion generation algorithm includes: Continuously generate n random numbers r1, r2, ..., r distributed in the range [0,1] n ; Determine the maximum switching frequency based on the switching frequency range of the current Markov chain state and minimum switching frequency ; By formula Calculate n random numbers r1, r2, ..., r respectively n The corresponding n switching frequencies f1, f2, ..., f n ; By adjusting the n switching frequencies f1, f2, ..., f n Perform weighted averaging to obtain the final switching frequency .

[0009] Furthermore, the random frequency conversion generation algorithm further includes: Based on the final switching frequency Determine the PWM frequency and PWM period; The action sequence of the basic vectors is arranged according to the existing symmetrical seven-segment SVPWM modulation method, and a random amount δ is added to the action time of the first zero vector V0 to make the pulse on time vary randomly.

[0010] Furthermore, the n random numbers r1, r2, ..., r n They are all generated by a hardware random number generator, which generates a random signal by collecting thermal noise or clock jitter in the circuit, and amplifies, filters, and quantizes the random signal to generate a random number distributed in the range of [0,1].

[0011] Furthermore, the n switching frequencies f1, f2, ..., f n Perform weighted averaging to obtain the final switching frequency ,include: Based on the stability of the working condition, the n switching frequencies f1, f2, ..., f n The corresponding weight values ​​ω1, ω2, ..., ω n ; In response to the stable working condition, the weight values ​​ω1, ω2, ..., ω n The values ​​of are increased in sequence; in response to the drastic change in working conditions, the weight values ​​ω1, ω2, ..., ω n The values ​​of are the same; By formula Perform weighted averaging and obtain the final switching frequency .

[0012] Furthermore, the method further comprises: Get the motor speed and torque requirements; In response to the motor speed not being within the preset speed range and the torque demand not being within the preset torque demand range, not starting random frequency conversion; In response to the motor speed being within a preset speed range and / or the torque demand being within a preset torque demand range, random frequency conversion is started.

[0013] Furthermore, the method further includes: issuing a fault alarm signal in response to data abnormality of the operating condition parameter.

[0014] A switching frequency switching system based on a Markov chain in the present invention includes: A working condition parameter acquisition module, which is used to obtain working condition parameters and update working condition parameters; a state definition module storing a predefined Markov chain and a working condition-state mapping rule; wherein the Markov chain includes a plurality of Markov chain states, each of which has an independent switching frequency interval and working condition parameter boundary; the working condition-state mapping rule states that if all working condition parameters fall within the range defined by the working condition parameter boundary of a certain Markov chain state, then the working condition is determined to correspond to the Markov chain state; A state matching module is used to determine a Markov chain state that matches the operating condition parameters based on a predefined Markov chain and operating condition-state mapping rule; A state transition probability matrix module, which has a pre-built state transition probability matrix and is used to obtain the state transition probability based on the Markov chain state before and after re-determination and the state transition probability matrix. It is also used to make state transition judgments based on the comparison results of the state transition probability and a preset threshold; and a frequency conversion module, which is used to generate a random switching frequency within a switching frequency interval of a determined Markov chain state; and is also used to regenerate the random switching frequency based on the transferred Markov chain state.

[0015] Furthermore, the state transition probability matrix module is further used for: Preset initial state transition probability matrix; Set the probability matrix update cycle, and based on the vehicle operation data, count the actual number of transitions and the total number of transitions for each Markov chain state in the previous probability matrix update cycle, and re-determine the state transition probability of each Markov chain state; Based on the re-determined state transition probabilities, the state transition probability matrix is ​​updated.

[0016] Furthermore, the frequency conversion module uses a random frequency conversion generation algorithm to generate a random switching frequency, and the frequency conversion module includes: A hardware random number generator is used to continuously generate n random numbers r1, r2, ..., r distributed in the range [0,1]. n ; A value determination unit for determining the maximum switching frequency based on the switching frequency interval of the current Markov chain state and minimum switching frequency ; Calculation unit, which is used to calculate the formula Calculate n random numbers r1, r2, ..., r respectively n The corresponding n switching frequencies f1, f2, ..., f n And by controlling n switching frequencies f1, f2, ..., f n Perform weighted averaging to obtain the final switching frequency , and based on the final switching frequency Determine the PWM frequency and PWM period; And a random pulse position pulse width modulation unit, which is used to arrange the action order of basic vectors according to the existing symmetrical seven-segment SVPWM modulation method, add a random amount δ to the action time of the first zero vector V0, and make the pulse conduction time randomly change.

[0017] Furthermore, the value obtaining unit is further configured to: determine the n switching frequencies f1, f2, ..., f based on the working condition stability. n The corresponding weight values ​​are ω1, ω2, ..., ω n ; In response to the stable working condition, the weight values ​​ω1, ω2, ..., ω n The values ​​of are increased in sequence; in response to the drastic change in working conditions, the weight values ​​ω1, ω2, ..., ω n The values ​​of are the same; The calculation unit is further used to: Perform weighted averaging and obtain the final switching frequency .

[0018] Furthermore, the system further includes a random frequency conversion start-up judgment module, which is used to: Get the motor speed and torque requirements; In response to the motor speed not being within the preset speed range and the torque demand not being within the preset torque demand range, not starting random frequency conversion; In response to the motor speed being within a preset speed range and / or the torque demand being within a preset torque demand range, random frequency conversion is started.

[0019] Furthermore, the system further includes a fault diagnosis module, which is configured to issue a fault alarm signal in response to data abnormality of the operating condition parameters.

[0020] A vehicle in the present invention includes the above-mentioned Markov chain-based switching frequency switching system.

[0021] The beneficial effect of the present invention is that after turning on random frequency conversion, the present invention realizes dynamic and adaptive adjustment of the vehicle's switching frequency by constructing a Markov chain model containing multiple specific states, combining an innovative random frequency conversion generation method, and combining a periodically updated state transition probability matrix, thereby significantly improving the comprehensive performance of the vehicle's power electronic system under different working conditions, and effectively balancing key performance indicators such as system efficiency, electromagnetic interference and dynamic response. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to make the purpose, technical solutions and beneficial effects of the present invention more clear, the present invention provides the following drawings for illustration: Figure 1 Schematic diagram of the flow of the switching frequency switching method based on the Markov chain in the first embodiment of the present invention; Figure 2 is the Markov chain state transition diagram in the first embodiment of the present invention; Figure 3 Arrange the action sequence diagram of basic vectors for the existing symmetrical seven-segment SVPWM modulation method; Figure 4 A sequence diagram showing the arrangement of basic vectors for the RPPSVPWM modulation method in the first embodiment of the present invention; Figure 5 Schematic diagram of the architecture of a switching frequency switching system based on a Markov chain in the second embodiment of the present invention. DETAILED DESCRIPTION

[0023] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0024] Example 1: like Figure 1 As shown, a switching frequency switching method based on a Markov chain in this embodiment includes: Step 1: Determine whether to enable random frequency conversion. This step specifically includes: Step 101, obtaining the motor speed and torque requirements; the motor speed and torque requirements are obtained in a conventional manner, for example, through a motor controller and / or a vehicle controller.

[0025] Step 102: In response to the motor speed not being within the preset speed range and the torque demand not being within the preset torque demand range, random frequency conversion is not enabled, and the current switching frequency control mode is maintained. In response to the motor speed being within the preset speed range and / or the torque demand being within the preset torque demand range, random frequency conversion is enabled, and then the process proceeds to step S2. In this step, the preset speed range may be 0-10,000 rpm, and the preset torque demand may be 0-200 N·m.

[0026] Step 2. After random frequency conversion is enabled, the operating condition parameters are obtained, and based on a predefined Markov chain and an operating condition-state mapping rule, a Markov chain state that matches the operating condition parameters is determined; wherein the Markov chain includes multiple Markov chain states, each of which has an independent switching frequency interval and operating condition parameter boundary; the operating condition-state mapping rule means that if all operating condition parameters fall within the range defined by the operating condition parameter boundary of a certain Markov chain state, then it is determined that the current operating condition corresponds to the Markov chain state.

[0027] The Markov chain can have five states, or any other number. When the predefined Markov chain includes five Markov chain states, the five Markov chain states are state S1, state S2, state S3, state S4, and state S5, and each of the Markov chain states has an independent switching frequency range and operating condition parameter boundary.

[0028] Operating condition parameters are collected via various sensors on the vehicle (such as current sensors, speed sensors, and voltage sensors). These parameters include, but are not limited to, load current, motor speed, battery voltage, and vehicle speed. The operating condition-state mapping rule states that each Markov chain state has corresponding operating condition parameter boundaries for different parameters. The collected operating condition parameters are compared with the operating condition parameter boundaries of each Markov chain state. If all operating condition parameters fall within the range defined by the operating condition parameter boundaries of a particular Markov chain state, the operating condition is determined to correspond to that Markov chain state.

[0029] Take five Markov chain states as an example: State S1: corresponding switching frequency range f min1 -f max1 (4KHz-5KHz), suitable for electric vehicles under light load conditions (such as empty vehicles or carrying a small number of passengers) and with extremely strict requirements on electromagnetic interference. For example, when driving in electromagnetically sensitive areas such as hospitals and communication base stations, the lower switching frequency range can effectively reduce electromagnetic radiation and meet strict electromagnetic compatibility requirements.

[0030] State S2: corresponding to the switching frequency range f min2 -f max2 (5KHz-6KHz) is suitable for electric vehicles with light loads but certain requirements for dynamic response. For example, when the vehicle performs frequent acceleration and deceleration operations under light load, this frequency range enables the system to respond quickly to load changes and ensure the stability of power output.

[0031] State S3: corresponding to the switching frequency range f min3 -f max3(6KHz-8KHz) is suitable for normal load conditions of electric vehicles, that is, when the vehicle carries a normal number of passengers and cargo and travels at a constant speed on urban roads or highways, this frequency range can take into account both system efficiency and stability.

[0032] State S4: corresponding to the switching frequency range f min4 -f max4 (8KHz-12KHz), suitable for working conditions where electric vehicles are heavily loaded (such as fully loaded with passengers and cargo) and have high dynamic response requirements. For example, when the vehicle is climbing a slope or accelerating to overtake, a higher switching frequency helps improve the dynamic performance of the system and ensure sufficient power output.

[0033] State S5: corresponding to the switching frequency range f min5 -f max5 (12KHz-15KHz), suitable for heavy-load conditions of electric vehicles with high efficiency requirements, such as when vehicles are transporting heavy loads over long distances. Through reasonable frequency adjustment, it can reduce switching losses while meeting heavy-load requirements and improve the overall efficiency of the system.

[0034] Step 3: Generate a random switching frequency using an algorithm within the switching frequency range of the determined Markov chain state.

[0035] Step 4: Update the operating parameters and redetermine the Markov chain state. Obtain the state transition probability based on the Markov chain state before and after redetermination and the pre-built state transition probability matrix.

[0036] When the predefined Markov chain contains five Markov chain states, the Markov chain state transition diagram is as follows: Figure 2 As shown: Pij represents the probability of transitioning from state Si to state Sj (e.g. P 23 represents the probability of transitioning from state S2 to state S3), satisfying , i=1, 2, 3, 4, 5. That is: the probability P of state S1 transitioning to state S1, state S2, state S3, state S4, state S5 11 、P 12 、P 13 、P 14 、P 15 The sum is 1; the probability P of state S2 transitioning to state S1, state S2, state S3, state S4, state S5 21 、P 22 、P 23 、P 24 、P 25 The sum is 1; the probability P of state S3 transitioning to state S1, state S2, state S3, state S4, and state S5 31 、P 32 、P33 、P 34 、P 35 The sum is 1; the probability P of state S4 transitioning to state S1, state S2, state S3, state S4, state S5 41 、P 42 、P 43 、P 44 、P 45 The sum is 1; the probability P of state S5 transitioning to state S1, state S2, state S3, state S4, and state S5 51 、P 52 、P 53 、P 54 、P 55 The sum is 1. The probabilities of mutual conversion between the above five states form a state transition probability matrix with five rows and five columns.

[0037] Updating the operating parameters and re-determining the Markov chain state means re-acquiring the operating parameters periodically at certain intervals and updating the Markov chain state according to the changes in the operating conditions. At the same time, the state transition probability can be obtained based on the Markov chain states before and after the re-determination and the pre-constructed state transition probability matrix. Due to the update of the operating parameters, the Markov chain state corresponding to the operating parameters is also updated based on the operating condition-state mapping rule. Therefore, the Markov chain state before the operating parameter update and the Markov chain state after the operating parameter update are both determined. The state transition probability from the Markov chain state before the update to the Markov chain state after the update can be directly obtained from the state transition probability matrix. The state transition probability obtained by this acquisition method is a definite value. The state transition probability can be used to compare with the preset threshold in the subsequent Step 5 to make a state transition judgment. For example, if the Markov chain state before the update is state S1 and the Markov chain state after the update is state S2, the state transition probability P can be obtained according to the state transition probability matrix. 12 The state transition probability P 12 It can be used to compare with the preset threshold in the subsequent Step 5 to make state transition judgment.

[0038] When the vehicle leaves the factory, an initial state transition probability matrix is ​​preset in the vehicle, and the state transition probability matrix is ​​updated based on the vehicle operation data generated while the user is driving the vehicle.

[0039] When the vehicle leaves the factory, the preset initial state transfer probability matrix in the vehicle means: using statistical analysis methods, through a large amount of real vehicle data, calculating the actual proportion of transfer times between states, and forming the initial state transfer probability matrix of the Markov chain.

[0040] When updating the state transition probability matrix, a probability matrix update cycle is pre-set at the factory or set by the user (for example, once every 5 hours). The actual number of transitions and the total number of transitions of each Markov chain state within the previous probability matrix update cycle are counted, and the state transition probability of each Markov chain state is re-determined. For example, within 5 hours of the previous probability matrix update cycle, the actual number of transitions from state S1 to state S1, state S2, state S3, state S4, and state S5 are 20 times, 30 times, 40 times, 50 times, and 60 times respectively. The total number of transitions of state S1 is 200 times. By calculating the ratio of the various actual transitions of state S1 to the total number of transitions of state S1 to be 10%, 15%, 20%, 25%, and 30% respectively, the updated probability P can be obtained. 11 、P 12 、P 13 、P 14 、P 15 They are 10%, 15%, 20%, 25%, and 30% respectively. Similarly, the values ​​of the elements in the remaining rows and columns of the state transition probability matrix can be updated.

[0041] Step 5: Based on the comparison between the state transition probability and the preset threshold, a state transition is determined. If the state transition occurs, a random switching frequency is regenerated based on the Markov chain state after the transition. If the state transition does not occur, the random switching frequency is not regenerated and the existing switching frequency is maintained.

[0042] Specifically, if the state transition probability obtained in Step 4 is greater than a preset threshold, the state transition is performed, and the random switching frequency is regenerated based on the Markov chain state after the transition and the random frequency conversion generation algorithm. If the state transition probability obtained in Step 4 is not greater than the preset threshold, the state transition is not performed.

[0043] Step 4 and Step 5 can be performed periodically to achieve adaptive switching of the switching frequency according to the real-time operating parameters of the vehicle.

[0044] Whether to enable random frequency conversion is determined based on the motor operating parameters. After enabling random frequency conversion, a Markov chain model containing multiple specific states is constructed, combined with an innovative random frequency conversion generation method and a periodically updated state transition probability matrix to achieve dynamic and adaptive adjustment of the vehicle's switching frequency, thereby significantly improving the comprehensive performance of the vehicle's power electronic system under different operating conditions and effectively balancing key performance indicators such as system efficiency, electromagnetic interference and dynamic response.

[0045] The "generating random switching frequency" in Step 3 and the "regenerating random switching frequency" in Step 5 both use a random frequency conversion generation algorithm. The steps of the random frequency conversion generation algorithm in Step 3 and Step 5 are the same, and the random frequency conversion generation algorithm includes the following steps: Step 3 / 501, continuously generate n random numbers r1, r2, ..., r distributed in the range [0,1] n .

[0046] The n random numbers r1, r2, ..., r n Both are generated using a hardware random number generator (HRG). This generator generates a random signal by sampling thermal noise or clock jitter in the circuit. This signal is amplified, filtered, and quantized to produce a random number distributed within the range [0, 1]. This HRG generates high-quality random numbers, avoiding the periodic repeatability issues associated with traditional pseudo-random number generation algorithms and ensuring the randomness and unpredictability of the switching frequency.

[0047] Step 3 / 502: Determine the maximum switching frequency based on the switching frequency range of the current Markov chain state and minimum switching frequency .

[0048] Step 3 / 503, through the formula Calculate n random numbers r1, r2, ..., r respectively n The corresponding n switching frequencies f1, f2, ..., f n .

[0049] Step 3 / 504, by adjusting the n switching frequencies f1, f2, ..., f n Perform weighted averaging to obtain the final switching frequency .

[0050] This step specifically includes: first, determining n switching frequencies f1, f2, ..., f based on the working condition stability. n The corresponding weight values ​​are ω1, ω2, ..., ω n Based on the stability of the working condition, it is to judge whether the working condition is stable based on the frequency of working condition changes. For example, within a probability matrix update cycle, if the total number of transitions of all Markov chain states is less than the preset number, the working condition is considered stable, and the weight values ​​ω1, ω2, ..., ω n The value of increases in turn, thereby increasing the weight of the recent random number. If the total number of transitions of all Markov chain states is greater than the preset number, it is considered that the working condition changes drastically, and the weight values ​​ω1, ω2, ..., ω nThe values ​​of are the same, and uniform weight distribution is adopted to make the frequency adjustment respond quickly to changes while avoiding excessive fluctuations. Then, through the formula Perform weighted averaging and obtain the final switching frequency .

[0051] Taking n=5 as an example, the five random numbers are: r1, r2, r3, r4, r5, and the corresponding five switching frequencies are: 、 、 、 、 If the working condition is stable at this time, the values ​​of ω1, ω2, ω3, ω4, and ω5 increase in sequence, then If the working conditions change drastically at this time, then ,Right now is the average of the five switching frequencies. ω1, ω2, ω3, ω4, and ω5 can be two arrays pre-stored in the frequency conversion module. One array contains ω1, ω2, ω3, ω4, and ω5 with increasing values ​​(for example, 2, 3, 4, 5, and 6, respectively), while the other array contains ω1, ω2, ω3, ω4, and ω5 with values ​​of 1. When performing the weighted average calculation of the switching frequency, the frequency conversion module selects one of the two arrays based on the stability of the operating conditions.

[0052] Calculate the final switching frequency by random numbers It is possible to generate random switching frequencies to ensure the randomness and unpredictability of the switching frequencies.

[0053] Step 3 / 505, based on the final switching frequency Determine the PWM frequency and PWM period. The PWM frequency of the existing symmetrical seven-segment SVPWM is equal to the final switching frequency , the PWM period is the inverse of the PWM frequency.

[0054] Step 3 / 506: Arrange the order of the basic vectors according to the existing symmetrical seven-segment SVPWM modulation method, and add a random value δ to the action time of the first zero vector V0 to randomly vary the pulse on-time. The random value δ can also be obtained using a hardware random number generator or a traditional random number generation algorithm.

[0055] like Figure 3 As shown in Figure 1, the switching state vector sequence of the symmetrical seven-segment SVPWM (space vector pulse width modulation) within one cycle is V0(000)→V4(100)→V6(110)→V7(111)→V6(110)→V4(100)→V0(000). The action time of the zero vector V0 in the first half of a PWM cycle is T.00 And the action time T of the zero vector V0 in the second half 66 Same, both .like Figure 4 As shown, after adding the random amount δ to the action time of the first zero vector V0, the action time T of the first half of the zero vector V0 00 becomes , the zero vector V0 of the second half , implement RPPSVPWM (Random PulsePosition SVPWM, random pulse position space vector pulse width modulation), so that the pulse on time changes randomly, that is, the pulse position changes randomly in one PWM cycle. The final switching frequency obtained in Step 3 / 504 It is already random. On this basis, random pulse position space vector pulse width modulation is used to further randomize the switching frequency, increasing randomness and unpredictability.

[0056] When obtaining and updating the operating parameters in Step 2 and Step 4, fault diagnosis is also performed to detect whether the data of the operating parameters (load current, motor speed, battery voltage, vehicle speed, etc.) are abnormal. When data abnormality is detected, an alarm signal is issued in time to remind the user to take corresponding protective measures to ensure the safe operation of the system.

[0057] Example 2: like Figure 5 As shown, a Markov chain-based switching frequency switching system in this embodiment is used to implement the Markov chain-based switching frequency switching method described in the first embodiment above. The system includes: an operating condition parameter acquisition module, a state definition module, a state matching module, a state transition probability matrix module, and a frequency conversion module. Each module is described below.

[0058] The operating condition parameter acquisition module is used to obtain and update operating condition parameters.

[0059] The operating condition parameter acquisition module collects operating condition parameters using various sensors on the vehicle (such as current sensors, speed sensors, and voltage sensors). These parameters include but are not limited to load current, motor speed, battery voltage, and vehicle speed. Updating operating condition parameters involves periodically reacquiring these parameters at regular intervals.

[0060] A state definition module stores a predefined Markov chain and an operating condition-state mapping rule; wherein the Markov chain includes multiple Markov chain states, each of which has an independent switching frequency interval and operating condition parameter boundary; the operating condition-state mapping rule means that if all operating condition parameters fall within the range defined by the operating condition parameter boundary of a certain Markov chain state, then it is determined that the current operating condition corresponds to the Markov chain state.

[0061] The number of predefined Markov chain states in the state definition module may be five, or any other number. When the predefined Markov chain includes five Markov chain states, the five Markov chain states are state S1, state S2, state S3, state S4, and state S5, and each Markov chain state has an independent switching frequency range and operating condition parameter boundary.

[0062] The working condition-state mapping rule means that corresponding working condition parameter boundaries are set for different parameters in each Markov chain state. By comparing the collected working condition parameters with the working condition parameter boundaries of each Markov chain state, if all working condition parameters fall within the range defined by the working condition parameter boundaries of a certain Markov chain state, it is determined that the working condition at this time corresponds to the Markov chain state.

[0063] Take five Markov chain states as an example: State S1: corresponding switching frequency range f min1 -f max1 (4KHz-5KHz), suitable for electric vehicles under light load conditions (such as empty vehicles or carrying a small number of passengers) and with extremely strict requirements on electromagnetic interference. For example, when driving in electromagnetically sensitive areas such as hospitals and communication base stations, the lower switching frequency range can effectively reduce electromagnetic radiation and meet strict electromagnetic compatibility requirements.

[0064] State S2: corresponding to the switching frequency range f min2 -f max2 (5KHz-6KHz) is suitable for electric vehicles with light loads but certain requirements for dynamic response. For example, when the vehicle performs frequent acceleration and deceleration operations under light load, this frequency range enables the system to respond quickly to load changes and ensure the stability of power output.

[0065] State S3: corresponding to the switching frequency range f min3 -f max3 (6KHz-8KHz) is suitable for normal load conditions of electric vehicles, that is, when the vehicle carries a normal number of passengers and cargo and travels at a constant speed on urban roads or highways, this frequency range can take into account both system efficiency and stability.

[0066] State S4: corresponding to the switching frequency range f min4 -f max4 (8KHz-12KHz), suitable for working conditions where electric vehicles are heavily loaded (such as fully loaded with passengers and cargo) and have high dynamic response requirements. For example, when the vehicle is climbing a slope or accelerating to overtake, a higher switching frequency helps improve the dynamic performance of the system and ensure sufficient power output.

[0067] State S5: corresponding to the switching frequency range f min5 -f max5 (12KHz-15KHz), suitable for heavy-load conditions of electric vehicles with high efficiency requirements, such as when vehicles are transporting heavy loads over long distances. Through reasonable frequency adjustment, it can reduce switching losses while meeting heavy-load requirements and improve the overall efficiency of the system.

[0068] The state matching module is used to determine the Markov chain state that matches the working condition parameters based on the predefined Markov chain and working condition-state mapping rules.

[0069] The state transition probability matrix module has a pre-built state transition probability matrix, which is used to obtain the state transition probability based on the Markov chain state before and after re-determination and the state transition probability matrix. It is also used to make state transition judgments based on the comparison results of the state transition probability and the preset threshold.

[0070] When the predefined Markov chain contains five Markov chain states, Pij represents the probability of transitioning from state Si to state Sj (e.g., P 23 represents the probability of transitioning from state S2 to state S3), the state transition probability matrix in the state transition probability matrix module satisfies , i=1, 2, 3, 4, 5. That is: the probability P of state S1 transitioning to state S1, state S2, state S3, state S4, state S5 11 、P 12 、P 13 、P 14 、P 15 The sum is 1; the probability P of state S2 transitioning to state S1, state S2, state S3, state S4, state S5 21 、P 22 、P 23 、P 24 、P 25 The sum is 1; the probability P of state S3 transitioning to state S1, state S2, state S3, state S4, and state S5 31 、P 32 、P 33 、P 34 、P 35The sum is 1; the probability P of state S4 transitioning to state S1, state S2, state S3, state S4, state S5 41 、P 42 、P 43 、P 44 、P 45 The sum is 1; the probability P of state S5 transitioning to state S1, state S2, state S3, state S4, and state S5 51 、P 52 、P 53 、P 54 、P 55 The sum is 1. The probabilities of mutual conversion between the above five states form a state transition probability matrix with five rows and five columns.

[0071] The frequency conversion module is used to generate a random switching frequency within the switching frequency range of the determined Markov chain state using a random frequency conversion generation algorithm; it is also used to regenerate the random switching frequency based on the transferred Markov chain state and the random frequency conversion generation algorithm.

[0072] Through the above-mentioned multiple modules, it is possible to achieve the following: after turning on random frequency conversion, by constructing a Markov chain model containing multiple specific states, combining an innovative random frequency conversion generation method, and combining a periodically updated state transition probability matrix, dynamic and adaptive adjustment of the vehicle's switching frequency can be achieved, thereby significantly improving the comprehensive performance of the vehicle's power electronic system under different operating conditions, and effectively balancing key performance indicators such as system efficiency, electromagnetic interference and dynamic response.

[0073] In this embodiment, the state transition probability matrix module is further used to: Preset initial state transition probability matrix; Set the probability matrix update cycle, and based on the vehicle operation data, count the actual number of transitions and the total number of transitions for each Markov chain state in the previous probability matrix update cycle, and re-determine the state transition probability of each Markov chain state; Based on the re-determined state transition probabilities, the state transition probability matrix is ​​updated.

[0074] When the vehicle leaves the factory, an initial state transition probability matrix is ​​preset in the vehicle, and the state transition probability matrix is ​​updated based on the vehicle operation data generated while the user is driving the vehicle.

[0075] When the vehicle leaves the factory, the preset initial state transfer probability matrix in the vehicle means: using statistical analysis methods, through a large amount of real vehicle data, calculating the actual proportion of transfer times between states, and forming the initial state transfer probability matrix of the Markov chain.

[0076] When updating the state transition probability matrix, a probability matrix update cycle is pre-set at the factory or set by the user (for example, once every 5 hours). The actual number of transitions and the total number of transitions of each Markov chain state within the previous probability matrix update cycle are counted, and the state transition probability of each Markov chain state is re-determined. For example, within 5 hours of the previous probability matrix update cycle, the actual number of transitions from state S1 to state S1, state S2, state S3, state S4, and state S5 are 20 times, 30 times, 40 times, 50 times, and 60 times respectively. The total number of transitions of state S1 is 200 times. By calculating the ratio of the various actual transitions of state S1 to the total number of transitions of state S1 to be 10%, 15%, 20%, 25%, and 30% respectively, the updated probability P can be obtained. 11 、P 12 、P 13 、P 14 、P 15 They are 10%, 15%, 20%, 25%, and 30% respectively. Similarly, the values ​​of the elements in the remaining rows and columns of the state transition probability matrix can be updated.

[0077] In this embodiment, the frequency conversion module includes: a hardware random number generator, a value obtaining unit, a calculation unit, and a random pulse position pulse width modulation unit.

[0078] A hardware random number generator is used to continuously generate n random numbers r1, r2, ..., r distributed in the range [0,1]. n The hardware random number generator generates a random signal by collecting thermal noise or clock jitter in the circuit. This signal is amplified, filtered, and quantized to produce a random number distributed within the range [0, 1]. This hardware random number generator generates high-quality random numbers, avoiding the periodic repeatability issues of traditional pseudo-random number generation algorithms and ensuring the randomness and unpredictability of the switching frequency.

[0079] A value determination unit for determining the maximum switching frequency based on the switching frequency interval of the current Markov chain state and minimum switching frequency . Maximum switching frequency and minimum switching frequency These are the upper and lower limits of the switching frequency range.

[0080] Calculation unit, which is used to calculate the Calculate n random numbers r1, r2, ..., r respectively n The corresponding n switching frequencies f1, f2, ..., f n And by controlling n switching frequencies f1, f2, ..., f nPerform weighted averaging to obtain the final switching frequency , and based on the final switching frequency Determine the PWM frequency and PWM period; the PWM frequency of the existing symmetrical seven-segment SVPWM is equal to the final switching frequency , the PWM period is the inverse of the PWM frequency.

[0081] The random pulse position pulse width modulation unit arranges the order of the basic vectors according to the existing symmetrical seven-segment SVPWM modulation method and adds a random amount δ to the duration of the first zero vector V0, causing the pulse on-time to vary randomly. The random amount δ can also be obtained through a hardware random number generator, or of course, using a traditional random number generation algorithm.

[0082] The switching state vector sequence of the symmetrical seven-segment SVPWM (space vector pulse width modulation) within one cycle is V0(000)→V4(100)→V6(110)→V7(111)→V6(110)→V4(100)→V0(000). The action time of the zero vector V0 in the first half of a PWM cycle is T. 00 And the action time T of the zero vector V0 in the second half 66 Same, both .like Figure 4 As shown, after adding the random amount δ to the action time of the first zero vector V0, the action time T of the first half of the zero vector V0 00 becomes , the zero vector V0 of the second half , implement RPPSVPWM (Random Pulse Position SVPWM, random pulse position space vector pulse width modulation), so that the pulse on time changes randomly, that is, the pulse position changes randomly in one PWM cycle. The final switching frequency obtained in Step 3 / 504 It is already random. On this basis, random pulse position space vector pulse width modulation is used to further randomize the switching frequency, increasing randomness and unpredictability.

[0083] In this embodiment, the value obtaining unit is further configured to: determine the n switching frequencies f1, f2, ..., f based on the working condition stability. n The corresponding weight values ​​are ω1, ω2, ..., ω n ; The calculation unit is further used to: Perform weighted averaging and obtain the final switching frequency .

[0084] Based on the stability of the working condition, it is to judge whether the working condition is stable based on the frequency of working condition changes. For example, within a probability matrix update cycle, if the total number of transitions of all Markov chain states is less than the preset number, the working condition is considered stable, and the weight values ​​ω1, ω2, ..., ω n The value of increases in turn, thereby increasing the weight of the recent random number. If the total number of transitions of all Markov chain states is greater than the preset number, it is considered that the working condition changes drastically, and the weight values ​​ω1, ω2, ..., ω n The values ​​of are the same and uniform weight distribution is adopted, so that the frequency adjustment can respond to changes quickly while avoiding excessive fluctuations.

[0085] Taking n=5 as an example, the five random numbers are: r1, r2, r3, r4, r5, and the corresponding five switching frequencies are: 、 、 、 、 If the values ​​of ω1, ω2, ω3, ω4, and ω5 increase in order, then If the working conditions change drastically at this time, then ,Right now is the average of the five switching frequencies. ω1, ω2, ω3, ω4, and ω5 can be two arrays pre-stored in the value-taking unit of the frequency conversion module. One array contains increasing values ​​for ω1, ω2, ω3, ω4, and ω5 (for example, 2, 3, 4, 5, and 6, respectively), while the other array contains values ​​for ω1, ω2, ω3, ω4, and ω5 all set to 1. When performing the weighted average calculation of the switching frequency, the calculation unit of the frequency conversion module selects one of the two arrays based on the stability of the operating conditions.

[0086] In this embodiment, the system further includes a random frequency conversion activation determination module, which is configured to: obtain motor speed and torque demand; disable random frequency conversion in response to the motor speed not being within a preset speed range and the torque demand not being within a preset torque demand range; and activate random frequency conversion in response to the motor speed being within the preset speed range and / or the torque demand being within a preset torque demand range. The preset speed range may be 0-10,000 rpm, and the preset torque demand may be 0-200 N·m.

[0087] In this embodiment, the system further includes a fault diagnosis module configured to issue a fault alarm signal in response to an abnormality in the operating parameter data. The fault diagnosis module detects abnormalities in the operating parameter data (load current, motor speed, battery voltage, vehicle speed, etc.). If an abnormality is detected, the module promptly issues an alarm signal to prompt the user to take appropriate protective measures to ensure safe system operation.

[0088] Example 3: A vehicle in the present invention includes the Markov chain-based switching frequency switching system in the second embodiment.

[0089] The vehicle may be, but is not limited to, a pure electric vehicle (Pure Electric Vehicle / Battery Electric Vehicle, PEV / BEV), a hybrid electric vehicle (Hybrid Electric Vehicle, HEV), a range extended electric vehicle (REEV), a plug-in hybrid electric vehicle (PHEV), a new energy vehicle (New Energy Vehicle), etc.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A switching frequency switching method based on a Markov chain, characterized in that: include: After random frequency conversion is enabled, operating condition parameters are obtained, and based on a predefined Markov chain and operating condition-state mapping rule, a Markov chain state that matches the operating condition parameters is determined; wherein the Markov chain includes a plurality of Markov chain states, each of which has an independent switching frequency interval and operating condition parameter boundary; the operating condition-state mapping rule is that if all operating condition parameters fall within a range defined by the operating condition parameter boundary of a certain Markov chain state, then it is determined that the current operating condition corresponds to the Markov chain state; generating a random switching frequency within the determined switching frequency interval of the Markov chain state; Updating the operating condition parameters and re-determining the Markov chain state, and obtaining a state transition probability according to the Markov chain state before and after the re-determination and a pre-constructed state transition probability matrix; A state transition judgment is performed based on a comparison result between the state transition probability and a preset threshold; if the state transitions, a random switching frequency is regenerated based on the Markov chain state after the transition.

2. The switching frequency switching method based on a Markov chain according to claim 1, characterized in that: The method further includes: constructing a state transition probability matrix, wherein constructing the state transition probability matrix specifically includes: Preset initial state transition probability matrix; Setting a probability matrix update period, and based on vehicle operation data, counting the actual number of transitions and the total number of transitions for each of the Markov chain states within the previous probability matrix update period, and re-determining the state transition probability for each of the Markov chain states; Based on the re-determined state transition probabilities, the state transition probability matrix is ​​updated.

3. The switching frequency switching method based on a Markov chain according to claim 1, characterized in that: The random switching frequency is generated by using a random frequency conversion generation algorithm, and the random frequency conversion generation algorithm includes: Continuously generate n random numbers r1, r2, ..., r distributed in the range [0,1] n ; Determine the maximum switching frequency based on the switching frequency range of the current Markov chain state and minimum switching frequency ; By formula Calculate n random numbers r1, r2, ..., r respectively n The corresponding n switching frequencies f1, f2, ..., f n ; By adjusting the n switching frequencies f1, f2, ..., f n Perform weighted averaging to obtain the final switching frequency .

4. The switching frequency switching method based on a Markov chain according to claim 3, characterized in that: The random frequency conversion generation algorithm also includes: Based on the final switching frequency Determine the PWM frequency and PWM period; The action sequence of the basic vectors is arranged according to the existing symmetrical seven-segment SVPWM modulation method, and a random amount δ is added to the action time of the first zero vector V0 to make the pulse on time vary randomly.

5. The switching frequency switching method based on a Markov chain according to claim 3, characterized in that: The n random numbers r1, r2, ..., r n They are all generated by a hardware random number generator, which generates a random signal by collecting thermal noise or clock jitter in the circuit, and amplifies, filters, and quantizes the random signal to generate a random number distributed in the range of [0,1].

6. The switching frequency switching method based on a Markov chain according to claim 3, characterized in that: The method is to adjust the n switching frequencies f1, f2, ..., f n Perform weighted averaging to obtain the final switching frequency ,include: Based on the stability of the working condition, the n switching frequencies f1, f2, ..., f n The corresponding weight values ​​ω1, ω2, ..., ω n ; In response to the stable working condition, the weight values ​​ω1, ω2, ..., ω n The values ​​of are increased in sequence; in response to the drastic change in working conditions, the weight values ​​ω1, ω2, ..., ω n The values ​​of are the same; By formula Perform weighted averaging and obtain the final switching frequency .

7. The switching frequency switching method based on a Markov chain according to claim 1, characterized in that: The method further comprises: Get the motor speed and torque requirements; In response to the motor speed not being within the preset speed range and the torque demand not being within the preset torque demand range, not starting random frequency conversion; In response to the motor speed being within a preset speed range and / or the torque demand being within a preset torque demand range, random frequency conversion is started.

8. The switching frequency switching method based on a Markov chain according to claim 1, characterized in that: The method further includes: issuing a fault alarm signal in response to data abnormality of the operating condition parameter.

9. A switching frequency switching system based on a Markov chain, characterized in that: include: A working condition parameter acquisition module, which is used to obtain working condition parameters and update working condition parameters; a state definition module storing a predefined Markov chain and a working condition-state mapping rule; wherein the Markov chain includes a plurality of Markov chain states, each of which has an independent switching frequency interval and working condition parameter boundary; the working condition-state mapping rule states that if all working condition parameters fall within the range defined by the working condition parameter boundary of a certain Markov chain state, then the working condition is determined to correspond to the Markov chain state; A state matching module is used to determine a Markov chain state that matches the operating condition parameters based on a predefined Markov chain and operating condition-state mapping rule; A state transition probability matrix module, which has a pre-built state transition probability matrix and is used to obtain the state transition probability based on the Markov chain state before and after re-determination and the state transition probability matrix. It is also used to make state transition judgments based on the comparison results of the state transition probability and a preset threshold; and a frequency conversion module, which is used to generate a random switching frequency within a switching frequency interval of a determined Markov chain state; and is also used to regenerate the random switching frequency based on the transferred Markov chain state.

10. The switching frequency switching system based on a Markov chain according to claim 9, characterized in that: The state transition probability matrix module is also used for: Preset initial state transition probability matrix; Set the probability matrix update cycle, and based on the vehicle operation data, count the actual number of transitions and the total number of transitions for each Markov chain state in the previous probability matrix update cycle, and re-determine the state transition probability of each Markov chain state; Based on the re-determined state transition probabilities, the state transition probability matrix is ​​updated.

11. The switching frequency switching system based on a Markov chain according to claim 10, characterized in that: The frequency conversion module uses a random frequency conversion generation algorithm to generate a random switching frequency, and the frequency conversion module includes: A hardware random number generator is used to continuously generate n random numbers r1, r2, ..., r distributed in the range [0,1]. n ; A value determination unit for determining the maximum switching frequency based on the switching frequency interval of the current Markov chain state and minimum switching frequency ; Calculation unit, which is used to calculate the formula Calculate n random numbers r1, r2, ..., r respectively n The corresponding n switching frequencies f1, f2, ..., f n And by controlling n switching frequencies f1, f2, ..., f n Perform weighted averaging to obtain the final switching frequency , and based on the final switching frequency Determine the PWM frequency and PWM period; and a random pulse position pulse width modulation unit, which is used to arrange the action order of the basic vectors according to the existing symmetrical seven-segment SVPWM modulation method, add a random amount δ to the action time of the first zero vector V0, and make the pulse conduction time randomly vary; The value obtaining unit is further configured to determine the n switching frequencies f1, f2, ..., f based on the working condition stability. n The corresponding weight values ​​are ω1, ω2, ..., ω n ; In response to the stable working condition, the weight values ​​ω1, ω2, ..., ω n The values ​​of are increased in sequence; in response to the drastic change in working conditions, the weight values ​​ω1, ω2, ..., ω n The values ​​of are the same; The calculation unit is further used to: Perform weighted averaging and obtain the final switching frequency .

12. A vehicle, characterized in that: It comprises a switching frequency switching system based on a Markov chain as described in any one of claims 9 to 11.

Images