Speed regulation control system of permanent magnet motor

Abstract

The invention relates to the technical field of permanent magnet motor speed regulation, in particular to a speed regulation control system of a permanent magnet motor, which realizes high-dimensional accurate description of the running state of the motor by identifying the magnetic performance change of a permanent magnet under different thermal state conditions and combining the structural change characteristics of a motor winding. Response consistency and error convergence performance of control gain adjustment are remarkably improved, a proportional integral factor value group is obtained by adjusting an interval gain sequence, ordered switching path planning between operation states is achieved, abnormal states are found in advance, an excitation fluctuation source is accurately positioned, and by constructing a voltage excitation adjustment set, the control gain adjustment accuracy is improved. The method effectively improves the flexible adjustment capability and the abnormity response rate of a control strategy, achieves the multi-time-domain, multi-segment, quick response and stable speed regulation of the permanent magnet motor under a complex working condition, remarkably improves the dynamic adaptive capability and the output control precision, and reduces the energy consumption and response lag.

Description

Technical Field

[0001] The present invention relates to the technical field of permanent magnet motor speed regulation, and particularly to a speed regulation control system for a permanent magnet motor. Background Art

[0002] The technical field of permanent magnet motor speed regulation aims at the technology of regulating the speed of a motor using a permanent magnet as the excitation source in electric drive and automation control systems. It has the advantages of compact structure, high efficiency, and large power density. It can dynamically adjust the output speed and torque of the motor according to the actual working conditions, and can operate efficiently and stably, meeting the system requirements of energy conservation, safety, and response speed.

[0003] The purpose of the speed regulation control system of a permanent magnet motor is to solve the problems of unstable speed regulation performance, high energy consumption, and response lag of a permanent magnet motor in scenarios of load fluctuation, starting response, low-speed operation, and high-precision positioning. Through a reasonable control strategy, it can achieve the effects of rapid motor startup, smooth operation, high-efficiency output, and wide-range speed regulation, improving the overall operation performance and intelligent level of the system.

[0004] Existing technologies generally rely on preset parameters based on static models and linear control strategies for speed regulation during operation. They are insufficient in responding to non-linear factors such as motor thermal state changes and electromagnetic feedback disturbances. When the load suddenly changes and there are frequent start-stop conditions, the control strategy is difficult to achieve instant perception and fine-grained adjustment of voltage and current disturbances, easily leading to control lag and overshoot in speed regulation, affecting the overall operation stability of the system. Existing technologies are difficult to accurately identify and specifically respond to abnormal excitation segments. When control instability and overshoot occur, they cannot be instantaneously corrected through refined gain partitioning and output amplitude regulation, severely restricting the performance release of the speed regulation system in scenarios such as intelligent manufacturing and precise motion control. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a speed regulation control system for a permanent magnet motor.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: A speed regulation control system for a permanent magnet motor includes:

[0007] A state recognition module: Based on the thermal behavior of the permanent magnet, the change characteristics of the motor winding structure, the magnitude of the magnetic flux feedback voltage, and the switching instruction flag bit, a set of operating state configurations is established;

[0008] A gain adjustment module: Based on the set of operating state configurations, a deep feedforward neural network is used to extract the synchronous interval of the speed increase and the error direction, construct a continuous error change interval, and compare it with the change trend of the jump amplitude of the current cycle gain factor to determine the relationship with the interval matching trend, and generate an adjustment interval gain sequence;

[0009] Path scheduling module: Based on the adjusted interval gain sequence, obtain the proportional-integral factor value group and the state node jump sequence, calculate the fitting of the temperature rise change trend and the average jump difference of the path segment, construct the control response delay sequence in combination with the response time of the continuous periodic error signal, extract the state node sequence according to the gain jump number, and at the same time construct the state transformation group diagram with the control response delay sequence to obtain the operation segment switching queue;

[0010] Risk monitoring module: Based on the operation segment switching queue, obtain the voltage amplitude difference between node segments and the stator current disturbance difference, extract the cross-cycle voltage amplitude mutation points and calculate the current drift peak deviation, and establish an excitation abnormal segment identification set;

[0011] Strategy switching module: Based on the excitation abnormal segment identification set, use the state segment number and the excitation direction of the control segment to adjust the corresponding control factor segment of the interval and select the section with the difference in output amplitude peak value, and construct a voltage excitation adjustment set.

[0012] As a further aspect of the present invention, the state recognition module includes:

[0013] Hot state sequencing sub-module: Based on the hot state behavior of the permanent magnet and the change characteristics of the motor winding structure, sample and sort the temperature points on the surface of the winding, segment and cut through the distance between the extreme temperature difference points to generate a hot zone distribution group, extract the extreme values of the response voltage amplitude in each hot zone to obtain the temperature difference change amplitude data, and then sort and construct a thermal fluctuation positioning index to generate a temperature zone mutation sequence;

[0014] Magnetic flux response sub-module: Based on the magnetic flux feedback voltage amplitude, construct the rotational speed change rate corresponding to the current transient response curve in the reverse jump section as the rotor response trend curve, and call the temperature zone mutation sequence generated in the hot state sequencing sub-module, extract the temperature zone number corresponding to the magnetic deviation section and align the response time window to obtain the thermal-magnetic combined mapping index information, and at the same time complete the recombination and matching of the magnetic deviation section and the thermal mutation section to generate a magnetic deviation structure set;

[0015] Operation mapping sub-module: Based on the temperature zone mutation sequence, the magnetic deviation structure set and the switching instruction flag bit, extract the voltage excitation command sequence in each deviation section and map the control segment flag, calculate the difference in the speed mutation amplitude before and after the extreme point and match the current jump trajectory section, then classify and merge the speed jump points and the excitation sections to obtain the operation state matching mapping and establish an operation state configuration set.

[0016] As a further aspect of the present invention, the gain adjustment module includes:

[0017] Synchronous extraction sub-module: Based on the set of operating state configurations, calculate the velocity difference on both sides of the velocity mutation point and generate a continuously changing vector. Through the velocity error signal sequence extracted from the set of operating state configurations, calculate the direction change condition as the direction change identifier, screen the matching sequence, and calculate the phase difference sequence based on the time sequence difference between the velocity signal and the excitation signal in the matching sequence. Divide the periodic segments with similar phase difference characteristics into phase groups, align the phase groups, use a deep feedforward neural network to extract the excitation signal mutation point, and perform intersection extraction with the overlapping region of the velocity direction switching point. Subsequently, exclude the offset jump segment region to generate a set of synchronous response segments;

[0018] Factor generation sub-module: Based on the set of synchronous response segments, generate the gradient of the error amplitude sequence within the same direction segment and extract the peak value of the jump ratio. Construct a fitting factor structure table by interpolating the initial values of the proportional and integral factors. Align the interpolation results according to the numbers of each jump segment in the synchronous response segment, calculate the ratio of the factor change before and after the jump to the increase in the periodic error amplitude to obtain the interpolation ratio of the jump segment. At the same time, extract the peak value of the control error corresponding to the period and perform difference cross-analysis with the parameter amplitude under the same period index in the fitting factor structure table to screen the effective parameters that meet the error constraint conditions and generate a set of excitation control factors;

[0019] Interval construction sub-module: Based on the set of excitation control factors, judge the direction consistency of the continuous control period factors and eliminate the samples with fluctuating reversals. Calculate the difference between the error mutation section and the fitting factor boundary to locate the section boundary, screen the fitting bandwidth of the continuous factor segments and merge them with the coincidence section verification to establish a sequence of adjustment interval gains.

[0020] As a further solution of the present invention, the path scheduling module includes:

[0021] Node extraction sub-module: Based on the sequence of adjustment interval gains, extract the proportional factor sequence, calculate and locate the extreme points of the jump amplitude difference in the proportional factor sequence. Extract the control period section index through the corresponding number of the integral factor and remove duplicates from the number sequence. Accumulate the length of the jump section and mark the continuous section number range to generate a set of state jump indexes;

[0022] Trend fitting sub-module: Based on the set of state jump indexes, match the corresponding section number range in the measured data of the surface temperature of the winding, and extract the temperature difference sequence within the corresponding section. Establish a fitting difference map by calculating the alignment method with the temperature increasing trend through the average change speed of the jump section, and extract the integration of the fitting error between sections and the average value of the change gradient to generate the result of the thermal delay change;

[0023] Structure construction sub-module: Based on the thermal delay change result, compare the excitation order between control section nodes and establish a path connection list, extract the peak value of the overlapping section through the coincidence ratio of the temperature rise change direction and time, rearrange the path combinations between nodes and construct a jump control mapping structure, and obtain the operation section switching queue.

[0024] As a further solution of the present invention, the rearrangement of the path combinations between nodes means obtaining the connection paths of multiple control section nodes, and then performing priority evaluation according to the coincidence degree between the temperature rise change slope corresponding to each path and the excitation time window. By rearranging the node order in the path from high to low according to the priority, the path segments that do not meet the time continuity and the temperature rise trend direction inconsistency are eliminated, and a control path sequence sorted by excitation consistency and thermal stability is generated.

[0025] As a further solution of the present invention, the risk monitoring module includes:

[0026] Amplitude detection sub-module: Based on the operation section switching queue, extract the voltage values between adjacent nodes and construct a voltage change sequence, calculate and identify the mutation jump points through the voltage difference between adjacent cycles and locate the section index, sort the voltage increase ratios of each section and extract the extreme values of the mutation sites, and generate a voltage mutation index group;

[0027] Perturbation identification sub-module: Based on the voltage mutation index group, extract the stator current waveforms within each mutation section and construct a cycle difference sequence, extract the current amplitude peak points through the fluctuation range, compare the historical average difference sequence, extract the fluctuation trend slope and the drift value of the mutation area, and generate a current drift feature set;

[0028] Abnormal annotation sub-module: Based on the current drift feature set, cross-screen the current drift peak positions and the voltage mutation section numbers and extract the overlapping number sequence, judge the abnormal interval of excitation persistence through the synchronous fluctuation of continuous sections, mark and integrate the state number association and the abnormal area, and establish an excitation abnormal section identification set.

[0029] As a further solution of the present invention, the strategy switching module includes:

[0030] Section number extraction sub-module: Based on the excitation abnormal section identification set, extract the excitation numbers in the section number sequence and classify and mark them according to the excitation direction symbols. Use the sections with opposite directions to exclude and the sections with the same direction to retain to generate an excitation vector group. After merging the continuous numbers, calculate the difference in the length of adjacent sections and screen the boundary sections to calibrate the jump demarcation points, and generate an excitation section number index set;

[0031] Factor screening sub-module: Based on the excitation segment number index set, extract the control factor segments corresponding to each segment number in the adjustment interval and construct a factor amplitude difference sequence. Calculate and extract the jump point peak value through the difference between the output control instruction change value and the factor mean value within the segment. Aggregate adjacent segment segments and generate a structured parameter matrix to generate an amplitude difference alignment structure group;

[0032] Instruction construction sub-module: Based on the amplitude difference alignment structure group, convert the index corresponding to the amplitude jump position of the control segment and generate a voltage control section sequence. Rewrite the control instruction values of each section and screen out non-jump sections. Replace the corresponding positions in the original control sequence while retaining the effective control segments, and splice them in chronological order to construct a complete control instruction set, and construct a voltage excitation adjustment set.

[0033] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0034] 1. In the present invention, by identifying the magnetic property changes of the permanent magnet under different thermal conditions and combining the characteristics of the motor winding structure changes, the high-dimensional accurate characterization of the motor operating state is realized, and the response consistency and error convergence performance of the control gain adjustment are significantly improved.

[0035] 2. In the present invention, by obtaining a proportional integral factor value group through the adjustment interval gain sequence and calculating the temperature rise change trend and the jump difference of the path segment in combination with the state node jump sequence, a state transformation diagram can be dynamically constructed to realize the orderly switching path planning between operating states, so as to discover abnormal states in advance and accurately locate the excitation fluctuation source.

[0036] 3. In the present invention, by constructing a voltage excitation adjustment set, the flexible adjustment ability and abnormal response rate of the control strategy are effectively improved, and the multi-time domain, multi-segment, fast response and stable speed regulation of the permanent magnet motor under complex working conditions are realized, significantly enhancing the dynamic adaptability and output control accuracy, and reducing energy consumption and response lag. Description of the Drawings

[0037] Figure 1 is the system flow chart of the present invention;

[0038] Figure 2 is the schematic diagram of the system framework of the present invention. Detailed Embodiments

[0039] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0040] Embodiment:

[0041] Please refer to Figure 1, the present invention provides a technical solution: A speed control system for a permanent magnet motor includes:

[0042] A state recognition module: Based on the thermal behavior of the permanent magnet, the change characteristics of the motor winding structure, the amplitude of the magnetic flux feedback voltage, and the switching instruction flag bit, a set of operating state configurations is established;

[0043] A gain adjustment module: Based on the set of operating state configurations, a deep feedforward neural network is used to extract the synchronization interval of the speed increase and the error direction, construct a continuous error change interval, and compare it with the change trend between the jump amplitudes of the current cycle gain factors to determine the relationship with the interval matching trend, and generate an adjusted interval gain sequence;

[0044] A path scheduling module: Based on the adjusted interval gain sequence, a set of proportional-integral factor values and a state node jump sequence are obtained, the fitting of the temperature rise change trend and the average jump difference of the path segment are calculated, a control response delay sequence is constructed in combination with the response time of the continuous cycle error signal, the state node sequence is extracted according to the gain jump number, and at the same time, a state transformation group diagram is constructed with the control response delay sequence to obtain an operating segment switching queue;

[0045] A risk monitoring module: Based on the operating segment switching queue, the difference between the voltage amplitude between node segments and the stator current disturbance is obtained, the voltage amplitude mutation points across cycles are extracted and the current drift peak deviation is calculated, and an excitation abnormal segment identification set is established;

[0046] A strategy switching module: Based on the excitation abnormal segment identification set, using the state segment number and the excitation direction of the control segment, the corresponding control factor segments of the adjustment interval are adjusted and the peak value of the output amplitude difference section is selected to construct a voltage excitation adjustment set.

[0047] Please refer to Figure 2 , the state recognition module includes:

[0048] A thermal sequencing sub-module: Based on the thermal behavior of the permanent magnet and the change characteristics of the motor winding structure, the surface temperature points of the winding are sampled and sorted, and the thermal zone distribution group is generated by segmenting and cutting through the distance between the extreme temperature difference points. The extreme values of the response voltage amplitude in each thermal zone are extracted to obtain the temperature difference change amplitude data, and then sorted to construct a thermal fluctuation positioning index to generate a temperature zone mutation sequence;

[0049] A magnetic flux response sub-module: Based on the amplitude of the magnetic flux feedback voltage, the corresponding rotational speed change rate in the current transient response curve within the reverse jump section is constructed as a rotor response trend curve, and the temperature zone mutation sequence generated in the thermal sequencing sub-module is called. The corresponding temperature zone number in the magnetic deviation section is extracted and the response time window is aligned to obtain the thermal-magnetic combined mapping index information, and at the same time, the recombination matching of the magnetic deviation section and the thermal mutation section is completed to generate a magnetic deviation structure set;

[0050] Operating mapping sub-module: Based on the temperature-region mutation sequence, the magnetic offset structure set, and the switching instruction flag bit, extract the voltage excitation command sequence within each offset section and map the control section flag, calculate the difference in the speed mutation amplitude before and after the extreme point and match the current jump trajectory section, then classify and merge the speed jump points and the excitation sections to obtain the operating state matching mapping, and establish the operating state configuration set;

[0051] Hot-state sequencing sub-module: Based on the thermal behavior of the permanent magnet, adopt the temperature distribution reconstruction method. Use the thermocouple array sensor to sample the temperature points on the surface of the winding every 0.5 seconds. Set the spatial distribution grid as a 10×10 matrix and use the center point of each cell as the sampling point position. The sampling data is spatially interpolated and fitted through the temperature matrix reconstruction function griddata(points, values, method='cubic') to extract the two-dimensional temperature field. Sort the temperature points using the quick_sort(array, low, high) of the quicksort algorithm in row-major order, record the sorted point numbers, and calculate the temperature difference vector ΔT for all point temperature values in sequence. Adopt the sliding extreme value analysis method with a window size of 5 points to extract the temperature difference change amplitude. Then perform pairwise difference operations on the indices of the extreme points in ΔT to calculate the extreme point spacing. Set the spacing threshold to 4 points for section segmentation to construct the hot zone distribution group. For the voltage response data corresponding to the sampling points in each hot zone, call the maximum and minimum functions np.max() and np.min() to extract the extreme values of the response voltage amplitude, and then call the argsort() function to sort the extreme point sequence to construct the thermal fluctuation positioning index and generate the temperature-region mutation sequence;

[0052] Magnetic flux response sub-module: Based on the temperature-region mutation sequence, adopt the magnetic flux change rate evaluation method. Use the difference function diff(V) to perform a first-order difference on the sampled magnetic flux voltage within each temperature region to construct the voltage change rate sequence. Use logical operators to judge the sign change pattern to screen the voltage direction reverse jump section. Call the conditional screening logic np.where(np.diff(np.sign(ΔV))!= 0) to select the sign switching point and extract the boundary index of the reverse section. Calculate the current change direction of the current frame and the previous frame in each reverse jump section, extract the current direction consistency section, and construct the voltage-current matching sequence through the Boolean matrix and the logical AND operation. Construct the magnetic field switching direction mapping table according to the indices in the matching sequence. The mapping table contains fields such as section number, jump direction, and current consistency value. Then associate the magnetic offset section position with each mapping item and merge and match it with the temperature mutation group according to the section number to generate the magnetic response section matrix containing the temperature, voltage, and current state vectors, and generate the magnetic offset structure set;

[0053] Operating mapping sub-module: Based on the magnetic offset structure set, using the operating state mapping modeling method, extract the voltage excitation command index sequence within the continuous time series of each magnetic offset section. Use the command ID field and timestamp field in the scheduling record data to construct and sort the excitation sequence. Call the groupby and np.diff() functions to calculate the interval of each excitation command. Map the control section flag field corresponding to the excitation command into the section index. Calculate the difference in speed measurement values within a 20ms window before and after the excitation command is issued. Use the central difference method Δv = (v[t + 1] - v[t - 1]) / 2 to calculate the speed mutation amplitude, where t represents the time index of the current analysis time point in the speed measurement sequence. Call the signal fitting function curve_fit() to fit the mutation trajectory and extract the change trend. Then call the second-order difference function np.diff(I, 2) of the current measurement sequence to extract the current jump trajectory section, where I represents the stator current measurement value sequence continuously recorded according to the sampling period. Classify and merge to obtain the speed jump point index and current mutation section, construct the excitation control section set, and use the logical aggregation function to match the one-to-one mapping relationship between speed, current, and excitation state to obtain the operating state matching mapping and establish the operating state configuration set.

[0054] Please refer to Figure 2 , the gain adjustment module includes:

[0055] Synchronous extraction sub-module: Based on the operating state configuration set, calculate the speed difference on both sides of the speed mutation point and generate a continuous change vector. Calculate the direction change situation as the direction change identifier through the speed error signal sequence extracted from the operating state configuration set. Screen and match the sequences, and calculate the phase difference sequence based on the time sequence difference between the speed signal and the excitation signal in the matching sequence. Divide the periodic segments with similar phase difference characteristics into phase groups and align the phase groups. Use a deep feedforward neural network to extract the excitation signal mutation point and perform an intersection extraction with the overlapping area of the speed direction switching point. Subsequently, exclude the offset jump segment area to generate a synchronous response section set;

[0056] Factor generation sub-module: Based on the synchronous response section set, generate the gradient of the error amplitude sequence within the same direction segment and extract the maximum jump ratio. Construct a fitting factor structure table by interpolating the initial values of the proportional and integral factors. Align the interpolation results according to the jump segment numbers in the synchronous response section. Calculate the ratio of the factor change before and after the jump to the increase in the periodic error amplitude to obtain the jump segment interpolation ratio. At the same time, extract the peak value of the control error corresponding to the period and perform a difference cross-analysis with the parameter amplitude under the same period index in the fitting factor structure table. Screen the effective parameters that meet the error constraint conditions to generate the excitation control factor set;

[0057] Interval construction sub-module: Based on the excitation control factor set, judge the consistency of the factor direction in consecutive control cycles and eliminate the samples with fluctuating reversals. Calculate the difference between the error mutation section and the fitting factor boundary to locate the section boundary. Screen the fitting bandwidth of the continuous factor segments and verify and merge them with the overlapping segments to establish an adjustment interval gain sequence;

[0058] Synchronous extraction sub-module: Based on the operating state configuration set, use the speed difference method to call np.diff(speed_sequence) to calculate the differences on both sides of the mutation point and generate a continuous change vector. Call np.sign(np.diff(error_signal)) to obtain the index of the direction change of the error signal and match the speed change sequence. Calculate the phase delay through the np.correlate() function and reconstruct the aligned sequence. Use a deep feedforward neural network to extract the overlapping region. The network structure includes three hidden layers, with 128 nodes in each layer. The activation function is ReLU, and the output layer is a linear structure. The input features are the instantaneous rotational speed, the estimated load, the first derivative of the error signal, and the change amplitude of the control signal in the previous cycle. Use model.predict([ω_t, T_est, de_dt, Δu]) to output the PI control factor and use it as an auxiliary criterion for the mutation point, where Δu is the change amplitude of the voltage control signal in two adjacent control cycles. Use np.intersect1d(trigger_index, switch_index) to extract the overlapping region of the excitation signal mutation point and the speed direction switching point. Introduce v as the time alignment error tolerance threshold to filter the abnormal matching pairs of offsets. Call np.where(abs(offset)<v) to exclude the offset skip section area and generate a synchronous response section set;

[0059] Factor generation sub-module: Based on the synchronous response section set, the piecewise gradient extraction method is used to calculate the error amplitude sequence. The np.diff(error_vector) function is called to perform the first-order difference on the data within the same-direction segment and the np.gradient() function is used to extract the local gradient value. The maximum jump ratio within each segment is extracted through the max(Δe) function, where Δe is the first-order difference sequence of the control error signal within the same-direction segment. The initial value intervals of the proportional and integral factors are constructed. The proportional factor interval is defined as Kp ∈ [0.1, 1.0], and the integral factor interval is Ki ∈ [0.01, 0.5]. The interp2d(Kp, Ki, Z) bilinear interpolation method is used to construct the factor structure table, where Z is the corresponding control error response result matrix under different combinations of the proportional factor Kp and the integral factor Ki. The interpolation ratio is matched using the principle that the difference between the fitting error within the jump segment and the interpolated output value in the structure table is minimized. The np.argmin(abs(fitted - actual)) is used to select the matching result, and the np.where() function is called to filter the effective parameter segments corresponding to the intersection of the control error peak value and the interpolation factor structure, generating the excitation control factor set;

[0060] Interval construction sub-module: Based on the excitation control factor set, the direction consistency judgment logic is used to calculate the factor direction difference within consecutive control periods. The np.where(Δθ == 0) is called to filter the segments with consistent directions and eliminate the samples with reversed directions, where Δθ represents the sequence of the control factor direction change within consecutive control periods. The absolute value of the difference between the error mutation section and the fitting factor boundary is calculated using the mutation boundary detection algorithm. The abs(error_edge - factor_edge) is called and a threshold of 0.05 is set to locate the corresponding section of the fitting boundary. The bandwidth aggregation analysis method is used to calculate the length of the continuous factor segments. The np.bincount() is called to count the continuous lengths and extract the segments longer than a specified value (such as N = 3). The np.intersect1d(segment_1, segment_2) is used to cross-compare the time series indices of the overlapping segments to verify and merge the factor bandwidth segments, establishing the adjustment interval gain sequence.

[0061] The deep feedforward neural network follows the formula: ;

[0062] Where: is the composite excitation mutation index value at the th sampling period, is the voltage mutation weight coefficient, is the voltage control signal value within the th control period, is the The voltage control signal value within one control period, is the absolute difference between adjacent periods of the voltage control signal, is the weight coefficient of the derivative of the error signal, is the first derivative of the error signal within the th control period, is the th error signal increment, is the sampling period duration, is the absolute value of the derivative of the error signal, is the weight coefficient of the proportional factor change value, is the proportional factor change value within the th period, is the absolute value of the proportional factor change value, is the weight coefficient of the load change rate, is the change rate of the load estimated value within the th period, is the load estimated value within the th period, is the absolute value of the load change rate. If is greater than the set threshold, then the determination point

[0063] Execution process: First, obtain the voltage control signal value of the current period and the value of the previous period , calculate the absolute difference , which is used as a basic measure of the voltage change amplitude. Extract the error value of the th period and the error value of the previous period from the system error signal sequence, calculate the derivative , that is, the error change rate. Then, output the proportional factor in the PI controller of the current period by the neural network structure and compare it with the output value of the previous period to obtain the proportional factor change , evaluate the self-regulation intensity of the controller. Obtain the load of the th period from the system load estimation module and calculate the derivative with the value of the previous period, which represents the change trend of the external working condition. Multiply the four measures by the set weight coefficients respectively, and sum them up with weights to obtain the composite excitation mutation index . Compare the value with the set threshold. When is greater than the set threshold, it is determined that there is an excitation signal mutation behavior in the current period, and the corresponding time point is included in the control parameter update logic and the synchronization matching sequence as a mutation point.

[0064] Please refer to Figure 2 , the path scheduling module includes:

[0065] Node extraction sub-module: Based on the adjustment interval gain sequence, a scale factor sequence is extracted, the extreme points of the jump amplitude difference in the scale factor sequence are calculated and located, the control period segment index is extracted through the corresponding number of the integral factor, the number sequence is de-duplicated and sorted, the length of the jump segment is accumulated, and the continuous segment number range is marked to generate a state jump index set;

[0066] Trend fitting sub-module: Based on the state jump index set, the corresponding segment number range in the measured data of the winding surface temperature is matched, and the temperature difference sequence in the corresponding section is extracted. A fitting difference map is established by calculating the average change speed of the jump section and aligning it with the temperature increasing trend. The fitting error integration and the average value of the change gradient between segments are extracted to generate the thermal delay change result;

[0067] Structure construction sub-module: Based on the thermal delay change result, the excitation order between the control section nodes is compared, and a path connection list is established. The maximum coincidence section is extracted through the coincidence ratio of the temperature rise change direction and time. The path combination between the nodes is rearranged, and a jump control mapping structure is constructed to obtain the operation section switching queue;

[0068] Node extraction sub-module: Based on the adjustment interval gain sequence, the extreme points of the scale factor sequence are calculated. The np.diff(Kp_sequence) function is called to calculate the difference sequence of adjacent scale factor values. After calculating the absolute value of the difference by np.abs(), the argrelextrema(array, np.greater) function is used to extract the index of the first-order maximum point as the index position of the extreme point of the jump amplitude difference. According to the corresponding number of the integral factor, the np.where(Ki_index == current_index) is used to extract the control period segment index set. The np.unique() function is called for the number sequence in the set to perform de-duplication and sorting. The loop structure is used to accumulate and judge the difference between consecutive numbers. If the number difference is 1, it is classified into the same segment. The length of the jump segment is accumulated, and the segment number interval range is defined. All segment numbers and extreme value indexes are combined to construct an index dictionary to generate a state jump index set;

[0069] Trend fitting sub-module: Based on the state jump index set, the window interval positioning method is used to match the temperature rise data number range, extract the temperature value sequence in the corresponding segment, calculate the temperature difference of each segment, construct the temperature difference change vector, and calculate the average temperature change rate by calling the ΔT / Δt formula through the time series of each segment in the jump segment. The linear trend matching algorithm is used to call the scipy.stats.linregress() function to fit the temperature change trend, obtain the fitting slope and intercept, use the difference between the output value of the fitting function and the actual temperature difference sequence as the fitting error, and then call np.mean(abs(fitted - actual)) to calculate the mean value of the change gradient. Construct a difference map structure array and generate the thermal delay change result;

[0070] Structure construction sub-module: Based on the thermal delay change result, the path connection matrix construction method is used. The excitation command order between the control segment nodes is used to extract the excitation timestamp sequence between each pair of control segments, and the excitation order list is constructed after sorting by calling np.argsort(command_time). A two-dimensional path connection array is established and initialized as a zero matrix. The excitation order pair is marked as the path start and end points, and the corresponding position in the matrix is assigned 1. The temperature rise change direction vector is cross-compared with the excitation time series, and the overlapping direction section is extracted by calling np.where(direction[i:j] == sign(ΔT[i:j])). Then, the maximum overlapping section index is extracted by using the time cross-ratio function np.intersect1d(time_list1, time_list2). The order between the path nodes is rearranged, and the node arrangement order is regenerated by the topological sorting algorithm topological_sort(path_list). A node jump connection dictionary is constructed, and then the path jump structure is assembled based on the index order and the jump control mapping structure is constructed to obtain the operation segment switching queue.

[0071] Rearranging the path combinations between nodes means obtaining the connection paths of multiple control segment nodes, and then evaluating the priority according to the coincidence degree between the temperature rise change slope corresponding to each path and the excitation time window. The node order in the path is rearranged from high to low according to the priority, and the path segments that do not meet the time continuity and the temperature rise trend direction inconsistency are removed, and a control path sequence sorted by excitation consistency and thermal stability is generated.

[0072] Please refer to Figure 2 , the risk monitoring module includes:

[0073] Amplitude detection sub-module: Based on the operation segment switching queue, the voltage values between adjacent nodes are extracted to construct a voltage change sequence. The mutation jump points are identified by calculating the voltage difference between adjacent cycles and the section index is located. The voltage increase ratios of each section are sorted and the extreme values of the mutation sites are extracted to generate a voltage mutation index group;

[0074] Perturbation recognition sub-module: Based on the voltage mutation index group, extract the stator current waveforms within each mutation segment and construct a period difference sequence. Extract the peak points of the current amplitude through the fluctuation range, compare with the historical average difference sequence, extract the fluctuation trend slope and the drift value in the mutation region, and generate a current drift feature set;

[0075] Abnormal annotation sub-module: Based on the current drift feature set, cross-screen the peak positions of the current drift and the voltage mutation point segment numbers and extract the coincidence number sequence. Judge the abnormal interval of the excitation persistence through the synchronous fluctuation of continuous sections, mark and integrate the state number association and the abnormal region, and establish an excitation abnormal segment identification set;

[0076] Amplitude detection sub-module: Based on the operation segment switching queue, use the differential analysis method to extract the voltage values between adjacent nodes. Call the loop structure to traverse the node number list, call U[i] and U[i+1] respectively to extract the corresponding voltage values and construct a sequence ΔU. Use np.diff(U_seq) to calculate the periodic voltage difference. Call the logical judgment np.where(abs(ΔU)>threshold) to identify the mutation jump points. Set the mutation judgment threshold to 20V. Extract the corresponding jump point numbers as the positioning index section according to the index. Call the sorting function np.argsort(abs(ΔU)) to sort the jump section indexes from largest to smallest in terms of the increase ratio. Extract the point with the largest absolute value among the first N jump sites, where N is set to 5. Store the extreme jump position numbers in the index list and generate a voltage mutation index group;

[0077] Perturbation recognition sub-module: Based on the voltage mutation index group, use the current differential fluctuation detection method to extract the stator current waveform data within the mutation segment. Call the sliding window method to extract the current values within each time series segment, with the window width set to 1 period, to generate a period difference sequence. Call the np.ptp(I_diff) function to calculate the range of the difference sequence and extract the peak points of the current amplitude. Compare each segment of the difference sequence with the historical average difference by calling np.mean(H_diff). Call np.polyfit(x, y, deg=1) for each segment sequence to perform a first-order fit to obtain the fluctuation trend slope, where x is the time index and y is the difference. Extract the difference between the start and end values of the current mutation segment to calculate the drift value in the mutation region drift = I[end] - I[start]. Construct a vector set containing the slope, drift value, and current range and generate a current drift feature set;

[0078] Abnormal annotation sub-module: Based on the current drift feature set, the index cross-screening method is used to match the current drift peak position with the voltage mutation segment number. Call np.intersect1d(current_peak_index, voltage_jump_index) to extract the overlapping number sequence. Extract the current and voltage sequences for the continuous time intervals in the overlapping segments. Call np.sign(diff(I)) == np.sign(diff(U)) to judge the consistency of the fluctuation direction in the continuous section. Count the continuous sections with a length exceeding the specified threshold and mark the segment numbers. Combine the abnormal numbers of each section with the original state numbers using zip() to construct a number-abnormal state dictionary structure and integrate and classify it, and establish an excitation abnormal segment identification set.

[0079] Please refer to Figure 2 , the strategy switching module includes:

[0080] Segment number extraction sub-module: Based on the excitation abnormal segment identification set, extract the excitation numbers in the segment number sequence and classify and mark them according to the excitation direction symbol. Use the segments with opposite directions to exclude and the segments with the same direction to retain to generate an excitation vector group. After merging the continuous numbers, calculate the difference in the lengths of adjacent segments and screen the boundary segments to calibrate the jump demarcation points, and generate an excitation segment number index set;

[0081] Factor screening sub-module: Based on the excitation segment number index set, extract the control factor fragments corresponding to each segment number and construct a factor amplitude difference sequence. Calculate and extract the maximum jump point through the difference between the output control instruction change value and the factor mean value within the fragment. Summarize the adjacent segment fragments and generate a structured parameter matrix, and generate an amplitude difference alignment structure group;

[0082] Instruction construction sub-module: Based on the amplitude difference alignment structure group, convert the index corresponding to the amplitude jump position of the control segment and generate a voltage control section sequence. Rewrite the control instruction values of each section and screen out the non-jump sections. Replace the corresponding positions in the original control sequence on the basis of retaining the effective control fragments, and splice them in chronological order to construct a complete control instruction set, and construct a voltage excitation adjustment set;

[0083] Segment number extraction sub-module: Based on the excitation anomaly segment identification set, the segment number sequence is processed using the classification label screening method. The np.array() function is called to convert the segment number sequence into an array structure. The np.sign(direction_sequence) function is used to generate a sign vector by using the excitation direction value corresponding to each segment number. The np.where() function is used to extract the segment numbers with positive direction signs as the retention objects. After excluding the reverse sign segments by boolean negation, an excitation vector group is constructed and the direction markers are recorded. The np.diff() function is called to calculate the difference between consecutive segment numbers for the numbers in the excitation vector group. The conditional judgment logic diff==1 is used to extract the set of consecutive numbers and the aggregation function is called for grouping and merging. The np.diff(length_vector) function is called to calculate the segment length difference for each group of segment numbers, and the np.argmax(abs(Δlength)) function is used to screen the boundary section with the largest change amplitude. The numbers and the index of the point with the largest difference are integrated to generate an excitation segment number index set;

[0084] Factor screening sub-module: Based on the excitation segment number index set, the factor difference construction method is used to extract the control factor segments in the adjustment interval. The index screening function segment_K=K_matrix[segment_index_list] is called to extract the corresponding factor segments from the original control factor matrix. The np.max() and np.min() functions are called for each segment of factor segments to calculate the difference and construct the amplitude difference sequence ΔK. The change value of the output control instruction ΔC=|C_out[i]-mean(K_segment)| is calculated and the np.argmax(ΔC) function is used to determine the point with the largest jump. The np.concatenate() function is called for the fragment sets of adjacent segments for data splicing. The merged fragments are called with the np.mean() and np.std() statistical functions to construct parameter matrix fields including mean, range, variance, and position index. The structured data of all segments are integrated to generate an amplitude difference alignment structure group;

[0085] Instruction construction sub-module: Based on the amplitude difference alignment structure group, the index mapping update method is used to perform a conversion operation on the amplitude jump positions of the control segments. The amplitude jump index list in the structure group is extracted. The position mapping function index_map[i]=control_index[i] is called in the original control segment array for each index to construct an index table. The voltage instruction corresponding to the segment number is extracted from the control segment instruction sequence according to the index. The instruction update function V_command[i]=new_value is called to rewrite the instruction value. The voltage values of the non-jump sections are screened using the jump identification vector. The full-interval segment number sequence is reconstructed and each segment of control fragments is called with np.append() for splicing and merging. It is organized into a standard control instruction structure array and a voltage excitation adjustment set is constructed.

[0086] The above are only the preferred embodiments of the present invention, and do not impose other forms of limitations on the present invention. Any person skilled in the relevant art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A speed control system for a permanent magnet motor, characterized in that, The system includes: A status recognition module: Based on the thermal behavior of the permanent magnet, the change characteristics of the motor winding structure, the amplitude of the flux linkage feedback voltage, and the switching instruction flag bit, a set of operating state configurations is established; A gain adjustment module: Based on the set of operating state configurations, a deep feedforward neural network is used to extract the synchronous interval of the speed increase and the error direction, construct a continuous error change interval, and compare it with the change trend of the jump amplitude of the current cycle gain factor to determine the relationship with the interval matching trend, and generate an adjusted interval gain sequence; A path scheduling module: Based on the adjusted interval gain sequence, a set of proportional-integral factor values and a state node jump sequence are obtained, the fitting of the temperature rise change trend and the average jump difference of the path segment are calculated, a control response delay sequence is constructed in combination with the response time of the continuous cycle error signal, the state node sequence is extracted according to the gain jump number, and at the same time, a state transformation group diagram is constructed with the control response delay sequence to obtain an operating segment switching queue; A risk monitoring module: Based on the operating segment switching queue, the voltage amplitude difference between node segments and the stator current disturbance difference are obtained, the cross-cycle voltage amplitude mutation points are extracted and the current drift peak deviation is calculated, and an excitation abnormal segment identification set is established; A strategy switching module: Based on the excitation abnormal segment identification set, the state segment number and the control segment excitation direction are used to adjust the corresponding control factor segments of the interval and select the segment with the difference from the output amplitude peak value to construct a voltage excitation adjustment set.

2. The speed control system of the permanent magnet motor according to claim 1, characterized in that The status recognition module includes: A thermal sequencing sub-module: Based on the thermal behavior of the permanent magnet and the change characteristics of the motor winding structure, the surface temperature points of the winding are sampled and sorted, the thermal zone distribution group is generated by segmenting through the distance between the extreme points of the temperature difference, the extreme values of the response voltage amplitude in each thermal zone are extracted to obtain the temperature difference change amplitude data, and then sorted to construct a thermal fluctuation positioning index and generate a temperature zone mutation sequence; A flux linkage response sub-module: Based on the amplitude of the flux linkage feedback voltage, the corresponding speed change rate in the current transient response curve in the reverse jump section is constructed as the rotor response trend curve, and the temperature zone mutation sequence generated in the thermal sequencing sub-module is called, the temperature zone number corresponding to the magnetic deviation section is extracted and the response time window is aligned to obtain the thermal-magnetic combined mapping index information, and at the same time, the recombination matching of the magnetic deviation section and the thermal mutation section is completed to generate a magnetic deviation structure set; An operating mapping sub-module: Based on the temperature zone mutation sequence, the magnetic deviation structure set, and the switching instruction flag bit, the voltage excitation command sequence in each deviation section is extracted and the control section mark is mapped, the speed mutation amplitude difference before and after the extreme point is calculated and the current jump trajectory section is matched, and then the speed jump points and the excitation sections are classified and merged to obtain an operating state matching mapping and establish a set of operating state configurations.

3. The speed control system of the permanent magnet motor according to claim 1, characterized in that, The gain adjustment module includes: Synchronous extraction sub-module: Based on the set of operating state configurations, calculate the velocity difference on both sides of the velocity mutation point and generate a continuously varying vector. Calculate the direction change situation as the direction change identifier through the velocity error signal sequence extracted from the set of operating state configurations, screen the matching sequences, and calculate the phase difference sequence based on the time sequence difference between the velocity signal and the excitation signal in the matching sequences. Divide the periodic segments with similar phase difference characteristics into phase groups, align the phase groups, use a deep feedforward neural network to extract the excitation signal mutation point, and perform intersection extraction with the overlapping region of the velocity direction switching point. Subsequently, exclude the offset jump segment region to generate a set of synchronous response segments; Factor generation sub-module: Based on the set of synchronous response segments, generate the gradient of the error amplitude sequence within the same-direction segment and extract the maximum jump ratio. Construct a fitting factor structure table through interpolation of the initial values of the proportional and integral factors. Align the interpolation results according to the jump segment numbers in the synchronous response segments, calculate the ratio of the factor change before and after the jump to the increase in the periodic error amplitude to obtain the interpolation ratio of the jump segment. At the same time, extract the peak value of the control error corresponding to the period, and perform difference cross-analysis with the parameter amplitude under the same period index in the fitting factor structure table to screen the effective parameters that meet the error constraint conditions and generate a set of excitation control factors; Interval construction sub-module: Based on the set of excitation control factors, judge the direction consistency of the continuous control period factors and eliminate the samples with fluctuating reversals. Calculate the difference between the error mutation section and the fitting factor boundary to locate the section boundary, screen the fitting bandwidth of the continuous factor segments and merge them with the coincidence section verification to establish an adjustment interval gain sequence.

4. The speed control system of the permanent magnet motor according to claim 1, characterized in that, The path scheduling module includes: Node extraction sub-module: Based on the adjustment interval gain sequence, extract the proportional factor sequence, calculate and locate the extreme points of the jump amplitude difference in the proportional factor sequence. Extract the control period segment index according to the corresponding number of the integral factor and remove duplicates from the number sequence. Accumulate the length of the jump section and label the continuous section number range to generate a set of state jump indices; Trend fitting sub-module: Based on the set of state jump indices, match the corresponding section number range in the measured data of the surface temperature of the winding, and extract the temperature difference sequence within the corresponding section. Establish a fitting difference map through the calculation of the average change speed of the jump section and the alignment method with the temperature increasing trend, and extract the integrated fitting error and the average value of the change gradient between each section to generate the thermal delay change result; Structure construction sub-module: Based on the thermal delay change result, compare the excitation order between the control section nodes and establish a path connection list. Extract the peak coincidence section through the coincidence ratio of the temperature rise change direction and time, rearrange the path combination between the nodes, and construct a jump control mapping structure to obtain the operation section switching queue.

5. The speed control system of the permanent magnet motor according to claim 4, characterized in that, The rearrangement of the path combination between nodes refers to obtaining the connection paths of multiple control section nodes, then evaluating the priority according to the coincidence degree of the temperature rise change slope corresponding to each path and the excitation time window. Rearrange the node order in the path from high to low according to the priority, and eliminate the path segments that do not meet the time continuity and the temperature rise trend direction inconsistency to generate a control path sequence sorted by excitation consistency and thermal stability.

6. The speed control system of the permanent magnet motor according to claim 1, characterized in that, The risk monitoring module includes: Amplitude detection sub-module: Based on the operation segment switching queue, extract the voltage values between adjacent nodes and construct a voltage change sequence. Identify mutation jump points by calculating the voltage differences between adjacent periods and locate the section index. Sort the voltage increase ratios of each section and extract the extreme values of the mutation sites to generate a voltage mutation index group; Disturbance identification sub-module: Based on the voltage mutation index group, extract the stator current waveforms within each mutation section and construct a period difference sequence. Extract the peak points of the current amplitude through the fluctuation range. Compare with the historical average difference sequence, and extract the fluctuation trend slope and the drift value in the mutation area to generate a current drift feature set; Abnormality annotation sub-module: Based on the current drift feature set, cross-screen the peak positions of the current drift and the section numbers of the voltage mutation points and extract the coincidence number sequence. Judge the abnormal interval of the excitation persistence through the synchronous fluctuation of consecutive sections, mark and integrate the state number association and the abnormal area, and establish an excitation abnormal section identification set.

7. The speed control system of the permanent magnet motor according to claim 1, characterized in that, The strategy switching module includes: Section number extraction sub-module: Based on the excitation abnormal section identification set, extract the excitation numbers in the section number sequence and classify and mark them according to the excitation direction symbols. Use the sections with opposite directions to exclude and the sections with the same direction to retain to generate an excitation vector group. After merging consecutive numbers, calculate the difference in the lengths of adjacent sections and screen the boundary sections to calibrate the jump demarcation points, and generate an excitation section number index set; Factor screening sub-module: Based on the excitation section number index set, extract the control factor fragments of the adjustment interval corresponding to each section number and construct a factor amplitude difference sequence. Extract the peak values of the jump points by calculating the difference between the output control instruction change value and the factor mean within the fragment. Summarize the fragments of adjacent sections and generate a structured parameter matrix to generate an amplitude difference alignment structure group; Instruction construction sub-module: Based on the amplitude difference alignment structure group, convert the index corresponding to the amplitude jump position of the control section and generate a voltage control section sequence. Rewrite the control instruction values of each section and screen out the non-jump sections. Replace the corresponding positions in the original control sequence on the basis of retaining the effective control fragments, and splice them in chronological order to construct a complete control instruction set to generate a voltage excitation adjustment set.

Images