Motor starting current limiting circuit, device based on soft starting cabinet and soft starting cabinet

Abstract

The application relates to the technical field of soft start cabinets, in particular to a motor starting current limiting circuit, a device and a soft start cabinet based on the soft start cabinet. In the circuit, a parameter calculation unit determines a reference current according to a motor rated current, a power grid voltage fluctuation value and a load type parameter, calculates a current overcurrent rate in combination with an actual starting current, and deduces and calculates a dynamic limiting coefficient based on a motor dynamic model. A data processing unit corrects a sampling area of sensor detection data according to the reference current and a soft start cabinet response delay, extracts effective detection data, and obtains an estimated limiting coefficient and a load type confidence degree through feature classification. A current control unit determines a target limiting coefficient by comprehensively considering the load type confidence degree, the calculated dynamic limiting coefficient and the estimated limiting coefficient, calculates a current limiting current in combination with the motor rated current and thyristor parameters, and finally limits and controls the output current of the soft start cabinet, so that the current suppression precision and the equipment reliability in the motor starting process are ensured.

Description

Technical Field

[0001] This invention relates to the field of soft starter cabinet technology, and in particular to a motor starting current limiting circuit, device and soft starter cabinet based on a soft starter cabinet. Background Technology

[0002] In the field of industrial motor control, soft starter cabinets are key devices for limiting motor starting current. They achieve smooth motor starting by gradually adjusting the output voltage, effectively avoiding damage to the power grid and motor caused by the current surge during direct starting. However, existing technologies typically rely on real-time detection data from a single current sensor for current limiting control. But under complex operating conditions, the instantaneous error in the detection data can easily lead to insufficient current limiting accuracy.

[0003] For example, when a motor drives a variable torque load such as a fan or pump, its starting current characteristics differ significantly from those of a constant torque load. A fixed current limiting algorithm struggles to balance dynamic response and current suppression under different load conditions. Furthermore, traditional current limiting methods often rely on preset fixed thresholds or simplified motor models, neglecting the dynamic changes in parameters such as motor impedance and back electromotive force during startup. This can easily lead to problems like over-protection causing excessively long startup times or under-protection causing current overruns. Especially in high-power motor startup scenarios, insufficient current limiting accuracy can result in overheating and damage to the thyristor module in the soft starter cabinet, or production interruptions due to frequent overcurrent protection triggers. Therefore, improving the current limiting accuracy of the soft starter cabinet by combining dynamic operating parameters with multi-source data fusion becomes a core issue in resolving the contradiction between current surges and equipment reliability during motor startup. Summary of the Invention

[0004] The main objective of this invention is to provide a motor starting current limiting circuit, device, and soft starter cabinet based on a soft starter cabinet, aiming to solve the technical problem of how to improve the current limiting accuracy of the soft starter cabinet by combining dynamic operating parameters and multi-source data fusion.

[0005] To achieve the above objectives, the present invention provides a motor starting current limiting circuit based on a soft starter cabinet, the circuit comprising: a parameter calculation unit, a data processing unit, and a current control unit;

[0006] The parameter calculation unit is connected to the data processing unit and the current control unit, respectively, and the current control unit is connected to the data processing unit and the soft starter cabinet, respectively.

[0007] The parameter calculation unit is used to determine the reference current based on the motor rated current, the grid voltage fluctuation value and the load type parameter, determine the current overcurrent rate based on the reference current and the actual starting current, and determine the dynamic limit coefficient based on the current overcurrent rate and the preset motor dynamic model.

[0008] The data processing unit is used to correct the sampling area of ​​the current sensor detection data according to the reference current and the soft starter cabinet response delay to obtain valid detection data, and to perform feature classification on the valid detection data to obtain the estimated limitation coefficient and load type confidence.

[0009] The current control unit is used to determine the target limiting coefficient based on the load type confidence level, the calculated dynamic limiting coefficient, and the estimated limiting coefficient, and to determine the current limiting current based on the target limiting coefficient, the motor rated current, and the thyristor parameters of the soft starter cabinet, and to limit and control the output current of the soft starter cabinet based on the current limiting current.

[0010] Optionally, determining the reference current based on the motor's rated current, grid voltage fluctuation value, and load type parameters includes:

[0011] The average reference current is determined based on the motor's rated current and the grid voltage fluctuation value, and the voltage fluctuation coefficient is determined based on the average reference current and the grid voltage fluctuation value.

[0012] The load correction factor is determined based on the load type parameters and the preset load type library;

[0013] The reference current is determined from the average reference current, the motor rated current, and the load correction current based on the voltage fluctuation coefficient and the load correction coefficient.

[0014] Optionally, determining the reference current from the average reference current, the motor rated current, and the load correction current based on the voltage fluctuation coefficient and the load correction coefficient includes:

[0015] In response to the voltage fluctuation coefficient being less than a preset first fluctuation threshold, the average reference current is determined as the reference current;

[0016] In response to the load correction factor being greater than or equal to a preset load factor threshold and the voltage fluctuation factor being greater than or equal to a preset first fluctuation threshold, a correction value of 1.2 times the rated current of the motor is determined as the reference current.

[0017] In response to the load correction coefficient being less than a preset load coefficient threshold, and the voltage fluctuation coefficient being greater than or equal to a preset first fluctuation threshold and less than a preset second fluctuation threshold, the weighted average of the average reference current and the load correction current is determined as the reference current.

[0018] In response to the load correction coefficient being less than a preset load coefficient threshold and the voltage fluctuation coefficient being greater than or equal to a preset second fluctuation threshold, a voltage correction current corresponding to the rated current of the motor is determined according to a preset voltage and current compensation model, and the voltage correction current is determined as the reference current.

[0019] Optionally, the step of correcting the sampling area of ​​the current sensor detection data based on the reference current and the soft starter cabinet response delay to obtain valid detection data includes:

[0020] Determine the trigger delay time of the current soft starter cabinet and the sampling window parameters of the sensor detection data; determine the current sampling delay compensation amount based on the reference current and the trigger delay time;

[0021] The current sampling window offset is determined based on the sampling window parameters and the load type parameters;

[0022] The sampling region is corrected on the time axis according to the sampling delay compensation amount and the sampling window offset to obtain the corrected sampling window, and the current sensor detection data is truncated according to the corrected sampling window to obtain the effective detection data.

[0023] Optionally, the step of performing feature classification on the valid detection data to obtain the estimated constraint coefficient and load type confidence score includes:

[0024] Feature extraction is performed on the effective detection data to obtain the current rise slope, steady-state current value, and temperature change rate;

[0025] The current rise slope, steady-state current value, and temperature change rate are input into a preset load classification model to obtain the load type and the load type confidence level corresponding to the effective detection data.

[0026] The retrieval coefficient corresponding to the load type is determined according to the preset type coefficient mapping relationship, and the retrieval coefficient is determined as the estimated constraint coefficient.

[0027] Optionally, determining the target limiting coefficient based on the load type confidence level, the calculated dynamic limiting coefficient, and the estimated limiting coefficient includes:

[0028] In response to the load type confidence level being less than a preset confidence threshold, the calculated dynamic constraint coefficient is determined as the target constraint coefficient;

[0029] In response to the load type confidence level being greater than or equal to a preset confidence threshold, the relative coefficient deviation is determined based on the calculated dynamic constraint coefficient and the estimated constraint coefficient;

[0030] In response to the relative coefficient deviation being less than a preset relative deviation threshold, the estimated constraint coefficient is determined as the target constraint coefficient;

[0031] In response to the relative coefficient deviation being greater than or equal to a preset relative deviation threshold, the weighted average of the calculated dynamic constraint coefficient and the estimated constraint coefficient is determined as the target constraint coefficient.

[0032] Optionally, determining the current limiting current based on the target limiting coefficient, the motor rated current, and the thyristor parameters of the soft starter cabinet includes:

[0033] The target coefficient current curve is determined from the preset coefficient current relationship library based on the rated current of the motor and the rated firing angle parameters of the thyristor of the soft starter cabinet.

[0034] The retrieved current value corresponding to the target limiting coefficient is determined based on the target coefficient current curve, and the retrieved current value is determined as the current limiting current.

[0035] Optionally, the step of limiting and controlling the output current of the soft starter cabinet based on the current limiting current includes:

[0036] In response to the soft starter cabinet's output current request value being less than or equal to the current limit current, the soft starter cabinet is controlled to adjust the thyristor firing angle according to the output current request value;

[0037] In response to the soft starter cabinet's output current request value being greater than the current limit current, the soft starter cabinet is controlled to adjust the thyristor firing angle according to the current limit current and issue an overcurrent warning signal.

[0038] In addition, to achieve the above objectives, the present invention also proposes a motor starting current limiting device based on a soft starter cabinet, wherein the motor starting current limiting device based on a soft starter cabinet includes the motor starting current limiting circuit based on a soft starter cabinet described above.

[0039] In addition, to achieve the above objectives, the present invention also proposes a soft starter cabinet, which includes the motor starting current limiting circuit based on the soft starter cabinet described above.

[0040] In this invention, the parameter calculation unit in the motor starting current limiting circuit based on the soft starter cabinet combines multi-dimensional parameters such as motor rated current, grid voltage fluctuations, and load type, breaking through the limitations of traditional single-sensor detection and making the determination of the reference current more closely match the actual working conditions. Based on the motor dynamic model, the dynamic limiting coefficient is calculated to respond in real time to changes in motor parameters during startup, avoiding over-protection or under-protection problems caused by fixed threshold algorithms. The data processing unit corrects the sensor sampling area for the response delay of the soft starter cabinet, filtering out invalid data caused by signal delay or noise, ensuring the timeliness and accuracy of the detection data. Through feature analysis... The system acquires load type confidence and estimated limit coefficients to effectively identify the starting current characteristics of variable torque loads such as fans and pumps, solving the problem of dynamic response lag when load changes suddenly under complex operating conditions. The current control unit integrates dynamic model calculations, data feature estimates, and load type confidence to form a multi-source fusion target limit coefficient, balancing the rigor of the theoretical model with the real-time performance of actual detection data. Combined with thyristor parameters, the limit current is dynamically adjusted, which can suppress the damage of starting current surges to the power grid and equipment, and avoid production interruptions caused by frequent overcurrent protection, significantly improving the reliability and stability of the soft starter cabinet in high-power motor starting scenarios. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the first embodiment of the motor starting current limiting circuit based on a soft starter cabinet according to the present invention;

[0042] Figure 2 This is a schematic diagram illustrating the specific process of the reference current confirmation step in the motor starting current limiting circuit based on the soft starter cabinet of this invention.

[0043] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0046] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0047] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0048] This invention provides a motor starting current limiting circuit based on a soft starter cabinet, referring to... Figure 1 As shown, Figure 1 This is a structural block diagram of the first embodiment of the motor starting current limiting circuit based on a soft starter cabinet according to the present invention. The motor starting current limiting circuit based on a soft starter cabinet of the present invention includes: a parameter calculation unit, a data processing unit, and a current control unit;

[0049] The parameter calculation unit is connected to the data processing unit and the current control unit, and the current control unit is connected to the data processing unit and the soft starter cabinet.

[0050] The parameter calculation unit is used to determine the reference current based on the motor rated current, grid voltage fluctuation value and load type parameters, determine the current overcurrent rate based on the reference current and the actual starting current, and determine the dynamic limit coefficient based on the current overcurrent rate and the preset motor dynamic model.

[0051] The data processing unit is used to correct the sampling area of ​​the current sensor detection data based on the reference current and the soft starter cabinet response delay to obtain valid detection data, and to perform feature classification on the valid detection data to obtain the estimated limit coefficient and load type confidence.

[0052] The current control unit is used to determine the target limit factor based on the load type confidence level, the calculated dynamic limit factor, and the estimated limit factor. It also determines the current limit current based on the target limit factor, the motor rated current, and the thyristor parameters of the soft starter cabinet, and controls the output current of the soft starter cabinet based on the current limit current.

[0053] The parameter calculation unit, connected electrically to the data processing unit and the current control unit, can generate an adaptive reference current based on the motor's rated current, grid voltage fluctuations, and load type parameters. It then calculates the current overcurrent rate by combining the actual starting current with the reference current and derives a dynamic limiting coefficient reflecting the changes in motor impedance and back electromotive force based on the motor's dynamic model. It is understandable that traditional current limiting control relies on fixed thresholds or simplified models, making it difficult to adapt to dynamic changes in motor parameters under different loads, easily leading to over-protection or under-protection.

[0054] In one exemplary embodiment, the parameter calculation unit may receive the motor's rated current, grid voltage fluctuation value, load type parameters, and actual starting current, generate a reference current through multi-dimensional operating condition parameter fusion, and use the motor dynamic model to solve for the limiting coefficient in real time to obtain the reference current, the current overcurrent rate, and the calculated dynamic limiting coefficient. Exemplarily, the parameter calculation unit may employ an embedded microcontroller to implement the algorithm calculation, or may include, but is not limited to, one or more methods using an FPGA-based hardware logic module for high-speed real-time calculation.

[0055] The data processing unit, electrically connected to the parameter calculation unit and the current control unit, can be used to correct sensor sampling area deviations caused by the soft starter cabinet's response delay, extract valid detection data, and obtain estimated constraint coefficients and load type confidence levels through pattern recognition of current waveform characteristics. It is understood that single sensor data is susceptible to noise interference and time delays; direct use can lead to delayed or misjudged control decisions, especially in scenarios with sudden load changes. In this embodiment, the data processing unit can receive sensor detection data, reference current, and soft starter cabinet response delay, compensate for system delays by dynamically adjusting the time window of the raw data, and use pattern recognition methods to classify and analyze the rise rate, peak shape, and harmonic components in the valid detection data to obtain valid detection data, estimated constraint coefficients, and load type confidence levels. Furthermore, the data processing unit may include a digital signal processor combined with filtering algorithms, or, but not limited to, integrating a machine learning inference engine for load feature identification, or one or more of these methods.

[0056] The current control unit, through electrical signal connections with the data processing unit, parameter calculation unit, and soft starter cabinet, can comprehensively calculate dynamic limiting coefficients, estimate limiting coefficients, and generate target limiting coefficients based on load type confidence levels. It then combines the motor's rated current and thyristor parameters to determine a safe and feasible current limiting current, thereby regulating the soft starter cabinet's output current. Understandably, traditional control relies solely on preset current limiting thresholds, failing to balance high-precision response with equipment safety. This can easily lead to thyristor overheating or frequent protection activation during high-power motor startup.

[0057] In one specific embodiment, the current control unit may receive calculated dynamic limiting coefficients, estimated limiting coefficients, load type confidence levels, motor rated current, and thyristor parameters; integrate model-driven and data-driven limiting suggestions through a weighted fusion strategy; and set the output current limit value based on the thyristor's safe operating boundary to obtain the current limit and control signal. For example, the current control unit may employ a PLC for logic control, or may use, but is not limited to, a dedicated PWM controller in conjunction with a drive circuit to perform one or more of the following: thyristor gate control.

[0058] It should be understood that the reference current is generated by the parameter calculation unit and transmitted to the data processing unit and its own internal modules. It can be used as a benchmark value to determine whether overcurrent has occurred during startup, and its value adaptively adjusts according to differences in motor rated current, grid voltage fluctuations, and load type. It is understandable that traditional solutions use a fixed multiple of the rated current as a reference, without considering grid fluctuations and load differences, leading to inaccurate startup control. In this embodiment, the reference current can be generated by receiving motor rated current, grid voltage fluctuation values, and load type parameters, and then using multivariate coupling modeling to generate a dynamic benchmark, outputting a benchmark current value for calculating the current overcurrent rate.

[0059] The motor's rated current, input to the parameter calculation unit, can be used to characterize the motor's standard current level under design operating conditions, serving as one of the fundamental static parameters for reference current modeling. In an exemplary embodiment, the motor's rated current can be derived from the motor nameplate or configuration information from the host system, serving as a key input for reference current calculation. The grid voltage fluctuation value, input to the parameter calculation unit, can reflect the real-time voltage deviation of the power supply network and participate in the correction of the expected current level during the startup phase.

[0060] Furthermore, the grid voltage fluctuation value can be real-time measurement data from voltage sensors or power monitoring modules, serving as a variable factor affecting the dynamic adjustment of the reference current. The load type parameter, input to the parameter calculation unit, can be used to identify whether the driven equipment belongs to a constant torque or variable torque load (such as a fan or pump), thereby affecting the starting characteristic modeling and dynamic model initialization. In a specific embodiment, the load type parameter can be a load category identifier from a manually set or automatically identified module, serving as the classification basis for the reference current and motor dynamic model construction.

[0061] The actual starting current is collected by a sensor and transmitted to the parameter calculation unit. It represents the real-time current flowing through the motor during startup, allowing for assessment of the current operating state and comparison with a reference current. For example, the actual starting current can be an analog or digital signal from a current transformer or Hall sensor, and its output is used to calculate the current overcurrent rate along with the reference current. The current overcurrent rate is generated within the parameter calculation unit and transmitted to the motor dynamic model calculation module. It can be used to quantify the excess ratio of the actual starting current relative to the reference current, serving as a trigger condition for adjusting the dynamic limit coefficient. In this embodiment, the current overcurrent rate can be calculated based on the ratio of the reference current to the actual starting current, and the output is a normalized overcurrent level index. The motor dynamic model is integrated within the parameter calculation unit and receives the current overcurrent rate as an excitation input. It can be used to describe the evolution of parameters such as stator impedance, rotor inertia, and back electromotive force over time during the motor startup phase, supporting the theoretical derivation of the dynamic limit coefficient. It is understood that traditional models struggle to reflect the nonlinear dynamic behavior under complex operating conditions.

[0062] Furthermore, the motor dynamic model can be constructed based on an improved equivalent circuit model of an asynchronous motor, or may include, but is not limited to, one or more methods that describe nonlinear dynamic behavior using state-space equations. The calculation of dynamic constraint coefficients, output from the parameter calculation unit to the current control unit, can be used to reflect the theoretically applicable current constraint strength derived from the motor dynamic model. In an exemplary embodiment, the calculation of dynamic constraint coefficients may involve receiving the computational results from the motor dynamic model as a model-driven component in the target constraint coefficient fusion process.

[0063] The sampling area is dynamically adjusted by the data processing unit based on the soft starter cabinet's response delay. This area defines the effective time interval for sensor data acquisition, and after delay compensation, ensures that the selected data reflects the current actual operating conditions. In this embodiment, the sampling area can be the time period for receiving the soft starter cabinet's response delay and the reference current change trend, outputting the corrected effective data. The soft starter cabinet's response delay, input to the data processing unit, can be used to characterize the time lag between issuing a control command and the actual output of the actuator, serving as the basis for sampling area correction. For example, the soft starter cabinet's response delay can be a delay time parameter from system calibration or online identification, outputting an offset for adjusting the sensor data sampling window. Sensor detection data from the current / voltage sensor is input to the data processing unit and can include real-time measurements such as raw current and voltage, forming the basic input source for data processing.

[0064] In one specific embodiment, the sensor detection data can be an analog or digitized electrical signal sequence, outputting as a raw data stream to be filtered and classified. Valid detection data is generated internally by the data processing unit and used by the feature classification module. It can represent high-quality detection data after denoising and time delay correction, supporting subsequent feature extraction and classification analysis. Furthermore, valid detection data can be received sensor detection data and sampling area correction results, outputting a clean dataset suitable for feature analysis.

[0065] Feature classification operates within the data processing unit. Inputting valid detection data, it outputs estimated constraint coefficients and load type confidence levels. It can be used for pattern recognition of features such as current rise rate, peak shape, and harmonic components to distinguish load types and estimate constraint requirements. Understandably, a lack of effective feature recognition will lead to inaccurate load identification. In an exemplary embodiment, feature classification may employ a support vector machine classifier, or one or more methods including, but not limited to, convolutional neural networks for end-to-end feature learning. The estimated constraint coefficients are output from the data processing unit to the current control unit and can be used to represent empirical constraint recommendations derived from sensor data feature analysis, reflecting the current regulation requirements under actual operating conditions.

[0066] In this embodiment, the estimated constraint coefficient can be the output of the received feature classification result, serving as the data-driven part of the target constraint coefficient fusion process. The load type confidence score, output from the data processing unit to the current control unit, can represent the reliability level of the current load type determination, and is a value between 0 and 1, used to adjust the weight allocation between the model and the data during the weighted fusion process. For example, the load type confidence score can be the matching degree score of the received feature classification module for the current load characteristics, output as a confidence index.

[0067] The target limiting coefficient is generated internally by the current control unit and used to calculate the current limiting current. It can represent the final decision value based on a combination of theoretical calculations and measured characteristics, determining the current limiting setting level. In one specific embodiment, the target limiting coefficient can be generated by receiving and calculating the dynamic limiting coefficient, the estimated limiting coefficient, and the load type confidence level, and then using a weighted fusion strategy to generate a unified control benchmark coefficient. The current limiting current is output by the current control unit to the soft starter cabinet actuator and can represent the final upper limit of the current applied to the soft starter cabinet. It is jointly determined by the target limiting coefficient, the motor rated current, and the thyristor parameters.

[0068] Furthermore, the current limiting current can receive the target limiting factor, the motor's rated current, and thyristor parameters, and output a specific current limiting command. The thyristor parameters are input to the current control unit and can be used to constrain the safe range of the current limiting current, including device characteristics such as on-state current, maximum surge capability, and thermal capacity.

[0069] In one exemplary embodiment, the thyristor parameters can be real-time parameters from the thyristor datasheet or a temperature feedback system, and the output is the safety boundary condition in the current limit current calculation. The soft starter cabinet output current is controlled by the soft starter cabinet itself, and its actual value is fed back to the sensor detection data link, which can be used to represent the actual current waveform supplied to the motor and is regulated by the current limit current command. In this embodiment, the soft starter cabinet output current can be the actual current waveform supplied to the motor, received from the current limit current command.

[0070] For example, in a scenario where a high-power water pump motor starts in a fluctuating power grid, the motor starting current limiting circuit based on a soft starter cabinet in this embodiment can be as follows: the power grid voltage fluctuation value is input to the parameter calculation unit, and a dynamic reference current is generated by combining the motor's rated current and known water pump load type parameters; the actual starting current is collected by a sensor to form detection data, and the data processing unit corrects the sampling area according to the response delay of the soft starter cabinet, eliminates hysteresis signals, and identifies the characteristics of typical variable torque load current curves through feature classification, outputting a high load type confidence level and corresponding estimated limiting coefficient; the parameter calculation unit simultaneously outputs a calculated dynamic limiting coefficient based on the motor dynamic model; the current control unit weights and fuses the two types of limiting coefficients according to the load type confidence level to obtain the target limiting coefficient, and adjusts the current limiting current in combination with the thyristor thermal parameters to achieve a smooth voltage boost output and avoid thyristor overload or protection malfunction.

[0071] This embodiment provides a motor starting current limiting circuit based on a soft starter cabinet. Through the coordinated connection and signal interaction of a parameter calculation unit, a data processing unit, and a current control unit, the parameter calculation unit integrates the motor's rated current, grid voltage fluctuation values, and load type parameters to generate an adaptive reference current. It then combines the actual starting current with the motor's dynamic model to derive and calculate the dynamic limiting coefficient, overcoming the control mismatch problem caused by traditional fixed thresholds and simplified models. The data processing unit utilizes the soft starter cabinet's response delay to correct the sensor sampling area, improving the timeliness and accuracy of the detection data. It also extracts load type confidence and estimated limiting coefficients through feature classification, enhancing the identification and response capabilities for variable torque loads. The current control unit comprehensively calculates the dynamic limiting coefficient, estimated limiting coefficient, and load type confidence to generate a target limiting coefficient. Combined with thyristor parameters, it determines a safe and controllable current limiting current, achieving precise control of the soft starter cabinet's output current. This effectively suppresses starting current surges, reduces the risk of thyristor overheating, and decreases the probability of overcurrent protection false triggering under complex operating conditions, significantly improving the reliability, stability, and success rate of the motor starting process.

[0072] In one embodiment, determining the reference current based on the motor's rated current, grid voltage fluctuation value, and load type parameters includes:

[0073] The average reference current is determined based on the motor's rated current and the grid voltage fluctuation value, and the voltage fluctuation coefficient is determined based on the average reference current and the grid voltage fluctuation value.

[0074] The load correction factor is determined based on the load type parameters and the preset load type library;

[0075] The reference current is determined from the average base current, the motor rated current, and the load correction current based on the voltage fluctuation coefficient and the load correction coefficient.

[0076] The average reference current is generated within the parameter calculation unit and participates in the determination of the reference current. It can be used as one of the candidate inputs for the reference current, based on the base current calculated from the motor's rated current and the grid voltage fluctuation, reflecting the impact of power supply voltage changes on the starting current. Understandably, traditional schemes only set the reference value as a multiple of the rated current, without considering the impact of grid voltage fluctuations on the actual starting current amplitude, leading to initial judgment errors.

[0077] In one exemplary embodiment, the average reference current can be obtained by receiving the motor's rated current and the grid voltage fluctuation value. A weighted average or function mapping operation is performed on the motor's rated current and the real-time grid voltage fluctuation value to form a reference current level adapted to the current power supply conditions, thus obtaining an intermediate variable for subsequent selection or adjustment of the reference current. Furthermore, the average reference current can use a linear proportional function to establish a voltage-current relationship model, or, including but not limited to, a lookup table method based on one or more different reference currents corresponding to voltage fluctuation intervals. The voltage fluctuation coefficient, derived from the grid voltage fluctuation value within the parameter calculation unit and applied to the reference current generation logic, can be used as a dimensionless coefficient to quantify the degree of influence of grid voltage fluctuations on the motor starting process, guiding the selection of the reference current composition.

[0078] Understandably, existing technologies lack explicit modeling of voltage disturbances and cannot dynamically adjust the reference current generation strategy to cope with weak power grids or voltage sag scenarios. In this embodiment, the voltage fluctuation coefficient can be obtained by receiving the grid voltage fluctuation value and calculating a value reflecting the system disturbance intensity based on the degree to which the grid voltage deviates from the rated value. This value serves as the decision basis for the multi-path selection mechanism, determining which of the average reference current, motor rated current, and load correction current should be selected as the reference current basis. For example, the voltage fluctuation coefficient may include a discrete selection signal output by a piecewise threshold comparator, or one or more methods including but not limited to constructing a continuous function to output weight coefficients for interpolation fusion.

[0079] The load correction factor is used within the parameter calculation unit and combined with the average reference current or other base current to adjust the final reference current. It can be used as a correction factor generated based on the matching results of the load type parameters and the preset load type library, reflecting the nonlinear influence of different types of loads (such as fans and pumps) on the starting current characteristics.

[0080] Understandably, traditional current limiting methods apply a uniform mode to all loads, failing to optimize the startup curve for the gradual increase characteristics of variable torque loads, resulting in energy waste or response lag. In a specific embodiment, the load correction coefficient can receive load type parameters and a preset load type library, and by querying the feature data in the preset load type library, convert the load type parameters into a proportional coefficient or gain value that can participate in the calculation, obtaining a multiplicative or additive coefficient used to correct the base current value. Furthermore, the load correction coefficient can include setting a fixed mapping table based on an empirical database, or including but not limited to introducing a fuzzy rule engine to achieve soft decision-making of the load type and adaptive generation of coefficients, or one or more of these methods. The preset load type library, built into the parameter calculation unit and called by the load correction coefficient module, can be used to store the startup characteristic parameter sets of various typical loads (such as constant torque, fans, water pumps, etc.), supporting the generation of load correction coefficients. Understandably, traditional control systems rely on manually setting the startup mode, lack standardized load characteristic support, and are difficult to achieve automatic adaptation.

[0081] In this embodiment, the preset load type library can receive load type parameters as indexes and output feature parameter templates for the corresponding load type by providing feature parameter templates for the corresponding load type. For example, the preset load type library may include a static lookup table structure, or one or more forms including but not limited to an updatable configuration file format for easy on-site maintenance and expansion.

[0082] The load correction current is generated within the parameter calculation unit and serves as one of the candidate inputs for the reference current. It can be used to adjust the current value based on the motor's rated current and the load correction coefficient, serving as a candidate form of reference current. In one exemplary embodiment, the load correction current can be achieved by receiving the motor's rated current and the load correction coefficient, and then performing a proportional correction through multiplication, or by using a polynomial function combined with multiple load characteristic dimensions for composite modeling to obtain the load-characteristic-compensated starting current reference. Further, the load correction current can include proportional correction through multiplication, or one or more methods including, but not limited to, using a polynomial function combined with multiple load characteristic dimensions for composite modeling.

[0083] For example, in scenarios where high-power wind turbines start under low-voltage conditions, the motor starting current limiting circuit based on the soft starter cabinet in this embodiment can be implemented by a parameter calculation unit receiving the motor's rated current, grid voltage fluctuation value, and load type parameters. First, it calculates the average reference current based on the rated current and voltage fluctuation value, and generates a corresponding voltage fluctuation coefficient. Simultaneously, it queries a preset load type library to confirm that wind turbine loads have a slowly increasing torque characteristic, and outputs a corresponding load correction coefficient. This coefficient is then used to correct the rated current, resulting in a corrected load current. Since the voltage fluctuation coefficient indicates a weak grid, the system selects the corrected load current as the primary current and further superimposes a voltage compensation term to ultimately determine the reference current. This method avoids the problems of insufficient starting torque due to insufficient voltage or applying excessively high initial current due to ignoring load characteristics, thus improving the starting success rate.

[0084] This embodiment determines a reference current based on the motor's rated current, grid voltage fluctuation, and load type parameters. This includes determining an average reference current based on the motor's rated current and grid voltage fluctuation, and determining a voltage fluctuation coefficient based on the average reference current and grid voltage fluctuation. A load correction coefficient is determined based on the load type parameters and a preset load type library. The reference current is then determined from the average reference current, motor rated current, and load correction current based on the voltage fluctuation coefficient and the load correction coefficient. A base current value reflecting the impact of supply voltage changes is calculated using the average reference current based on the motor's rated current and grid voltage fluctuation, and this base current is used as a candidate input for the reference current. The voltage fluctuation coefficient quantifies the degree of influence of grid voltage fluctuations on the startup process. The reference current serves as a control factor for selecting the path of the reference current. A correction factor reflecting the nonlinear characteristics of the load is generated based on the matching results between the load type parameters and the preset load type library using a load correction coefficient. The starting characteristic parameters of typical loads are stored in the preset load type library to support the generation of the correction coefficient. The load correction current, combined with the load correction coefficient, forms a current reference compensated for load characteristics. The coordinated operation of these structures ensures that the reference current not only reflects the rated characteristics of the equipment but also responds in real time to changes in the power grid status and load conditions. This significantly improves the accuracy of the initial decision-making in current limiting control, effectively avoids over-protection or under-protection problems caused by initial reference deviation, and enhances the adaptability and reliability of the soft starter cabinet in complex industrial environments.

[0085] Further, in one embodiment, determining a reference current from the average reference current, the motor rated current, and the load corrected current based on the voltage fluctuation coefficient and the load correction coefficient includes:

[0086] In response to the voltage fluctuation coefficient being less than a preset first fluctuation threshold, the average reference current is determined as the reference current;

[0087] In response to a load correction factor greater than or equal to a preset load factor threshold and a voltage fluctuation factor greater than or equal to a preset first fluctuation threshold, a correction value of 1.2 times the rated motor current is determined as the reference current.

[0088] In response to the load correction factor being less than a preset load factor threshold, and the voltage fluctuation factor being greater than or equal to a preset first fluctuation threshold and less than a preset second fluctuation threshold, the weighted average of the average reference current and the load correction current is determined as the reference current.

[0089] In response to a load correction factor less than a preset load factor threshold and a voltage fluctuation factor greater than or equal to a preset second fluctuation threshold, a voltage correction current corresponding to the motor rated current is determined according to a preset voltage and current compensation model, and the voltage correction current is determined as the reference current.

[0090] Among them, the first fluctuation threshold serves as a primary dividing standard for judging the degree of grid voltage fluctuation. It is used to distinguish between mild and moderate voltage disturbance scenarios. It participates in the condition judgment logic within the parameter calculation unit and is compared with the voltage fluctuation coefficient. It can be used to trigger a logic signal to determine whether to use the average reference current as the reference current.

[0091] Understandably, traditional systems lack an explicit mechanism for classifying voltage fluctuation levels, making it impossible to adjust control strategies based on disturbance intensity. In an exemplary embodiment, the first fluctuation threshold can be a preset numerical threshold configuration that outputs a corresponding judgment signal based on the comparison result with the voltage fluctuation coefficient. For example, the first fluctuation threshold can include, but is not limited to, one or more of the following: a normalized coefficient range based on ±5% of the rated voltage, or an empirical range calibrated through on-site commissioning. The second fluctuation threshold serves as the judgment boundary for high-voltage fluctuation conditions, used to identify severe grid disturbances and activate the compensation model. It is compared with the voltage fluctuation coefficient within the parameter calculation unit and participates in the reference current selection decision, and can be used to trigger a control signal to determine whether to enable the voltage-current compensation model. Existing technologies lack a multi-level voltage anomaly response mechanism, making it difficult to achieve refined current limiting control under strong disturbance conditions.

[0092] Furthermore, the second fluctuation threshold can be a preset numerical threshold configuration, which generates a high disturbance identification signal by comparing it with the voltage fluctuation coefficient at different levels. In one specific embodiment, the second fluctuation threshold may include, but is not limited to, one or more of the following: a normalized fluctuation level corresponding to ±10% or more of the rated voltage, or a safety limit derived from the thyristor withstand voltage capability.

[0093] The load factor threshold is used to determine the classification boundary of the significance of load characteristics, distinguishing between load types with high and low correction requirements. Within the parameter calculation unit, it is compared with the load correction factor and influences the selection of the reference current generation path. It can be used as a basis for determining whether to apply a multiple correction to the motor's rated current. Understandably, traditional methods lack a clear quantification threshold for load impact, leading to the application of the same starting mode to different types of loads, resulting in insufficient control precision.

[0094] In this embodiment, the load factor threshold can be a preset empirical threshold or a dynamically updated threshold obtained through self-learning. The load type identification result is output by comparing it with the load correction coefficient. For example, the load factor threshold can be, but is not limited to, a fixed proportional value (e.g., 0.7), or a dynamic threshold derived from cluster analysis of historical operating data, or one or more of these. The 1.2x correction value is the intermediate value obtained by multiplying the motor's rated current by 1.2. It serves as a candidate reference current for enhancing starting capability under specific operating conditions. It is generated within the parameter calculation unit and selected as the reference current when the conditions are met. It can be used to output the current value adjusted by 1.2x gain. It is understood that conventional soft starters only use a fixed multiple (e.g., 6x) for current limiting, which cannot optimize the starting margin for moderate disturbances and typical load combinations.

[0095] Furthermore, the 1.2x correction value can be an enhanced reference value adapted to the current operating condition obtained by receiving the rated current of the motor and amplifying it by a fixed ratio. In an exemplary embodiment, the 1.2x correction value may include, but is not limited to, one or more of the following: multiplication operations implemented by a fixed gain module, or dynamic adjustment of the multiplier by a programmable gain amplifier.

[0096] The weighted average is the result of fusing the average reference current and the load correction current according to their respective weights. It reflects the combined influence of both the power grid and the load. Calculated within the parameter calculation unit, it serves as the reference current value under specific conditions and can be used to output the intermediate current value that fuses the two types of correction terms. Understandably, traditional schemes fail to achieve coordinated fusion of multi-source current references, making it difficult to balance the control contradictions between power supply quality and load characteristics.

[0097] In this embodiment, the weighted average value can be generated by linearly combining the received average reference current, the load correction current, and a preset weighting coefficient to take into account multiple influences. For example, the weighted average value can be implemented using a linear weighting formula, or by introducing a fuzzy inference mechanism to dynamically adjust one or more of the weight allocation methods.

[0098] The voltage-current compensation model is a mathematical model describing the relationship between the degree of voltage anomaly and the required starting current. It is used to generate an appropriate reference current under extreme operating conditions. Built into the parameter calculation unit, it receives the motor's rated current and voltage fluctuation coefficient as input and outputs a voltage correction current. Its operating principle is based on the motor's equivalent circuit and the voltage-torque relationship, dynamically compensating for the starting current requirement under low or overvoltage conditions. Understandably, existing control systems lack active compensation mechanisms for severe voltage deviations, which can easily lead to starting failure or current runaway.

[0099] Furthermore, the voltage and current compensation model can receive the motor's rated current and voltage fluctuation coefficient, and solve for a highly adaptable compensation current value through a nonlinear mapping relationship. In a specific embodiment, the voltage and current compensation model can be constructed, but is not limited to, in a lookup table interpolation form, or using a nonlinear function expression to achieve a continuous mapping, or one or more of these methods. The voltage correction current is a reference current value output by the voltage and current compensation model that adapts to severe voltage fluctuations, ensuring effective starting capability even under extreme conditions. It is generated within the parameter calculation unit and determined as the reference current when the conditions are met, and can be used to provide one of the reference current options for final current limiting control.

[0100] Understandably, this design addresses the challenge of current adaptation under high-amplitude voltage disturbances. In this embodiment, the voltage-corrected current can be the calculation result of the received voltage-current compensation model, which is then converted into actual control commands through a digital or analog processing link. For example, the voltage-corrected current can be obtained, but is not limited to, through real-time model solving by a digital computing unit, or implemented by analog circuitry in conjunction with an ADC / DAC link, or one or more of these methods.

[0101] For example, in a scenario where a large water pump motor starts under complex conditions of a sudden rise in grid voltage and weakened load characteristics, the motor starting current limiting circuit based on the soft starter cabinet in this embodiment can detect that the voltage fluctuation coefficient exceeds the second fluctuation threshold, indicating a serious voltage anomaly; at the same time, the load correction coefficient is lower than the load coefficient threshold, reflecting that the current load characteristics are not obvious or deviate from the typical variable torque mode; the system determines that it has entered a high disturbance low load identification state, starts the preset voltage and current compensation model, dynamically calculates the voltage correction current based on the motor's rated current, and uses it as the reference current; this value fully considers the initial impact risk caused by the voltage rise, while avoiding excessive restriction or relaxation of the allowable current due to misjudgment of the load type, realizing safe and stable start-up under extreme conditions, and preventing thyristor damage or frequent protection operation due to instantaneous overcurrent.

[0102] This embodiment determines the reference current from the average reference current, motor rated current, and load correction current based on the voltage fluctuation coefficient and load correction coefficient. It includes multi-level judgment logic responding to different voltage fluctuation and load conditions. A first fluctuation threshold distinguishes minor voltage disturbances and triggers the selection of the average reference current. A load coefficient threshold identifies the significance of load characteristics to determine whether to enable the multiplier correction. A 1.2x correction value provides adequate starting margin when moderate disturbances and typical loads coexist. A weighted average value integrates both grid and load factors under moderate voltage fluctuations and weak load influences. A second fluctuation threshold identifies severe voltage anomalies and activates the voltage-current compensation model. This model dynamically outputs the voltage correction current based on the motor's equivalent relationship to adapt to extreme operating conditions. This achieves adaptive selection of the reference current under different grid conditions and load characteristic combinations, improving the adaptability of the current limiting strategy to complex operating environments, effectively suppressing overcurrent risks during high-power motor startup, reducing thyristor overheating and protection malfunctions, thereby improving the safety, stability, and control accuracy of the soft-start process.

[0103] In one embodiment, the sampling area of ​​the current sensor detection data is corrected based on the reference current and the soft starter cabinet response delay to obtain valid detection data. This includes: determining the trigger delay time of the current soft starter cabinet and the sampling window parameters of the sensor detection data; determining the current sampling delay compensation amount based on the reference current and the trigger delay time; determining the current sampling window offset based on the sampling window parameters and the load type parameters; correcting the sampling area on the time axis based on the sampling delay compensation amount and the sampling window offset to obtain a corrected sampling window; and truncating the current sensor detection data based on the corrected sampling window to obtain valid detection data.

[0104] The trigger delay time is input to the data processing unit to calculate the current sampling delay compensation. It can be used to characterize the time lag between the soft starter cabinet receiving the control signal and the actual conduction of the thyristor, thereby quantifying the system response delay and participating in sampling correction. It is understandable that traditional data sampling does not consider the actual delay of the thyristor trigger action, causing the sampling period to deviate from the true current change range, affecting control accuracy.

[0105] In one exemplary embodiment, the trigger delay time can be the thyristor trigger response delay obtained from system calibration or online detection, a fixed value obtained through factory static calibration, or dynamically updated using a real-time feedback mechanism, to obtain a time parameter used as the basis for calculating the current sampling delay compensation. The sampling window parameter is input to the data processing unit to participate in the generation of the current sampling window offset and the corrected sampling window. It can be used to define the basic time length and periodic characteristics of sensor data acquisition, serving as the basic configuration for determining the effective sampling range. It is understood that a fixed sampling window cannot adapt to changes in the current rise rate during different load startup processes and is prone to including invalid transient data. Furthermore, the sampling window parameter can be a timing reference obtained by receiving timing parameters such as the sampling frequency and window width set from the control system, using a fixed-period sliding window or a variable-length adaptive window setting, to jointly determine the sampling window adjustment strategy with the load type parameter.

[0106] The current sampling delay compensation is derived from the trigger delay time and reference current within the data processing unit and used to correct the sampling region. It can be used to calculate the time compensation value based on the reference current change trend and the trigger delay time, achieving forward correction of the sampling start point. It is understood that due to the thyristor turn-on delay, without forward compensation, the sampling interval will lag behind the actual current change process. In this embodiment, the current sampling delay compensation can be obtained by receiving the rate of change or phase information of the trigger delay time and the reference current, and by calculating the amount of time to be sampled earlier based on the expected current increase during the trigger delay period, thus obtaining the time shift applied to the original sampling region.

[0107] The current sampling window offset, determined by both the sampling window parameters and the load type parameters, is used within the data processing unit to adjust the sampling area. It reflects changes in the optimal sampling segment position due to load type differences, enabling adaptation to the starting characteristics of variable torque loads. Understandably, loads such as fans and pumps exhibit significantly different current rise rates during startup, making it difficult for a fixed sampling center to cover key characteristic intervals. For example, the current sampling window offset can receive the sampling window parameters and load type parameters, dynamically adjusting the data interval of interest based on the current rise rate characteristics of different loads (e.g., fans, pumps), thus obtaining an adjustment amount for the center or starting position of the sampling area. The sampling delay compensation amount participates in the construction of the corrected sampling window and can be used, together with the current sampling delay compensation amount, to achieve pre-correction on the time axis, ensuring that sampling covers key current change stages.

[0108] It is understood that the technical necessity has already been covered in the existing description, and this statement is retained here for the sake of consistency in terminology. In a specific embodiment, the sampling delay compensation amount can be the result of the correlation analysis between the received trigger delay time and the reference current, and a compensation value in the time dimension can be obtained through an operation mechanism consistent with the aforementioned delay compensation logic, which is then used for the overall translation operation of the subsequent sampling area.

[0109] For example, in the scenario of a high-inertia wind turbine motor under cold start conditions, the motor starting current limiting circuit based on the soft starter cabinet in this embodiment can be as follows: the data processing unit obtains the trigger delay time of the current soft starter cabinet in real time, and determines the basic sampling interval by combining it with preset sampling window parameters; the current sampling delay compensation amount is calculated based on the rise rate of the reference current and the trigger delay time, and the sampling start point is advanced; at the same time, based on the slow current rise characteristics of wind turbine loads, a larger current sampling window offset is obtained by adjusting the load type parameters; the delay compensation amount and the window offset are combined to form a corrected sampling window, and the current sensor detection data in the window is extracted to exclude the non-steady-state fluctuation part in the initial stage of startup, so as to obtain effective detection data representing the real operating state and improve the accuracy of subsequent limiting coefficient estimation.

[0110] This embodiment determines the trigger delay time and sampling window parameters of the current soft starter cabinet and sensor detection data. It determines the current sampling delay compensation based on the reference current and trigger delay time, and the current sampling window offset based on the sampling window parameters and load type parameters. Then, it corrects the sampling area's time axis based on the sampling delay compensation and sampling window offset to obtain the corrected sampling window. Based on this, it extracts the current sensor detection data to obtain valid detection data. The trigger delay time quantifies the system response lag and serves as the basis for calculating the current sampling delay compensation. It combines the reference current change rate to infer the amount of time that should be sampled earlier to achieve forward compensation. The sampling window parameters and load type parameters are used in conjunction to generate sampling window adjustment amounts for different load start-up characteristics. Finally, the two types of compensation amounts are combined to complete the overall time axis correction of the original sampling area. This achieves the technical effect of eliminating non-steady-state transients, noise interference, and delay distortion, extracting highly timely and representative valid detection data, improving the accuracy of current sensing, and enhancing the dynamic response capability and stability of the soft starter cabinet under complex operating conditions.

[0111] In one embodiment, feature classification is performed on the valid detection data to obtain estimated constraint coefficients and load type confidence scores, including:

[0112] Feature extraction is performed on the effective detection data to obtain the current rise slope, steady-state current value, and temperature change rate;

[0113] Input the current rise slope, steady-state current value, and temperature change rate into the preset load classification model to obtain the load type and load type confidence level corresponding to the valid detection data;

[0114] Based on the preset type coefficient mapping relationship, the retrieval coefficient corresponding to the load type is determined, and the retrieval coefficient is determined as the estimated constraint coefficient.

[0115] Feature extraction involves receiving valid detection data and analyzing time-series data using numerical differentiation, trend fitting, and state recognition methods to obtain representative dynamic indicators, such as the current rise slope, steady-state current value, and temperature change rate. It is understandable that raw sensor data contains redundant information and is subject to noise interference, making its direct use for classification decisions unreliable; therefore, feature reduction is necessary to improve recognition accuracy.

[0116] In one exemplary embodiment, feature extraction may include extracting the initial current change trend from valid detection data to calculate the current rise slope, identifying the current amplitude in the stabilizing phase to obtain the steady-state current value, and processing the real-time temperature sampling sequence to generate the temperature change rate. The current rise slope may be a quantitative parameter characterizing the rate of current growth during the initial startup of the motor, generated by the feature extraction module, and used as input to a preset load classification model.

[0117] For example, the current rise slope can be an output signal calculated from the initial current change trend in the effective detection data, based on the current increment per unit time. This parameter reflects the electrical inertial response characteristics of the load at startup and is one of the important electrical characteristics distinguishing between variable torque and constant torque loads. The steady-state current value can be a measurement parameter reflecting the current level of the motor before it enters stable operation, generated by the feature extraction module and input into the load classification model. Furthermore, the steady-state current value can be a current measurement value near the steady state extracted from the current amplitude in the stable phase of the effective detection data. This parameter can be used to assist in judging the load's impedance characteristics and mechanical inertia.

[0118] The rate of temperature change can be a thermodynamic indicator describing the rate of temperature rise of a motor or power device during startup. Real-time temperature sampling sequences collected by temperature sensors are processed and input into the feature extraction module. In one specific embodiment, the rate of temperature change can be an output signal obtained by analyzing the amount of temperature change per unit time, providing information about the load's thermal response behavior and enhancing the multi-dimensional discrimination capability during the classification process. The load classification model can receive features such as the current rise slope, steady-state current value, and rate of temperature change. It then performs pattern recognition on the input multi-dimensional feature vector using a pre-trained or configured classification algorithm to determine the current load category and provide classification confidence, thereby outputting the corresponding load type and load type confidence score.

[0119] Understandably, traditional load identification relies on manual settings or simple threshold judgments, which cannot adapt to dynamic changes in characteristics during startup, easily leading to a mismatch between the current limiting strategy and the actual load. In this embodiment, the input of the load classification model is connected to the feature extraction module, and the output is connected to the estimated limit coefficient generation logic, realizing automatic reasoning from feature input to load identification results. For example, the load classification model can use a decision tree model to achieve fast reasoning, or use a neural network model to improve classification accuracy under complex working conditions. The type coefficient mapping relationship can establish a lookup relationship between load type and corresponding limit coefficient, storing the suggested limit coefficient values ​​suitable for various loads through a preset table or function, realizing a direct mapping from load identification results to control parameters. Understandably, traditional current limiting strategies lack an effective mechanism to convert load identification results into specific control parameters, leading to a disconnect between perception and control. Furthermore, the input of the type coefficient mapping relationship is the load type output by the load classification model, and the output is the corresponding retrieval coefficient, which is connected to the estimated limit coefficient generation stage. In an exemplary embodiment, the type coefficient mapping relationship can include a static lookup table or be dynamically calculated using a parameterized function expression.

[0120] For example, in a scenario where a soft starter cabinet drives an unknown type of load to start, the motor starting current limiting circuit based on the soft starter cabinet in this embodiment can be achieved by the data processing unit receiving valid detection data, performing feature extraction to obtain a combination of features such as a low current rise slope, a gradually increasing steady-state current value, and a slow temperature change rate. These features are then input into a preset load classification model, which outputs that the current load is a "water pump" type variable torque load with high confidence. Subsequently, the model looks up the corresponding retrieval coefficient for the water pump load based on a preset type coefficient mapping relationship and passes it as an estimated limiting coefficient to the current control unit. This process achieves automatic identification of the load type and matching of the current limiting strategy without prior settings, improving the system's adaptability under non-standard operating conditions.

[0121] This embodiment extracts features from effective detection data to obtain the current rise slope, steady-state current value, and temperature change rate. The feature extraction module performs dimensionality reduction and noise reduction on the original data, extracting key dynamic indicators to accurately characterize the electrical and thermal behavior of the motor during startup. These features are input into a preset load classification model, and a classification algorithm intelligently identifies the load type and outputs the corresponding load type confidence level. This allows for a highly reliable distinction between variable torque loads (such as fans and pumps) and constant torque loads, solving the problem of current limiting strategy mismatch caused by load misjudgment. Based on a preset type coefficient mapping relationship, the corresponding retrieval coefficient is found based on the identified load type and used as the estimated limiting coefficient, achieving a closed-loop mapping from sensing results to control parameters. This enables the current limiting strategy to have adaptive adjustment capabilities for different load characteristics. This scheme transforms the multidimensional features of operating data into quantifiable control basis, enhancing the decision reliability of the data processing unit under complex operating conditions. It provides effective support for the current control unit to integrate dynamic limiting coefficients and estimated limiting coefficients, ultimately improving the current limiting accuracy and real-time response of the soft starter cabinet under diverse load conditions.

[0122] In one embodiment, determining the target limiting factor based on load type confidence, calculating the dynamic limiting factor, and estimating the limiting factor includes:

[0123] In response to a load type confidence level being less than a preset confidence threshold, the calculated dynamic limiting coefficient is determined as the target limiting coefficient.

[0124] In response to a load type confidence level greater than or equal to a preset confidence threshold, the relative coefficient deviation is determined based on the calculated dynamic limit coefficient and the estimated limit coefficient.

[0125] In response to a relative coefficient deviation being less than a preset relative deviation threshold, the estimated constraint coefficient is determined as the target constraint coefficient;

[0126] In response to a relative coefficient deviation being greater than or equal to a preset relative deviation threshold, the weighted average of the calculated dynamic constraint coefficient and the estimated constraint coefficient is determined as the target constraint coefficient.

[0127] The preset confidence threshold is input to the current control unit and participates in the logical condition judgment. It can be used as a judgment boundary to determine whether the load type identification result is reliable, and to trigger the branch selection of the target limit coefficient generation strategy. It is understood that traditional control lacks a mechanism for evaluating the reliability of perceived information and cannot automatically switch to a conservative safety strategy when uncertainty is high. In an exemplary embodiment, the preset confidence threshold can be a fixed or adjustable threshold parameter received from system configuration or an adaptive algorithm. By comparing it with the load type confidence level, a judgment signal is output to determine whether to enable the control path based on the estimated limit coefficient.

[0128] The preset confidence threshold may include, but is not limited to, one or more of the following: a fixed numerical threshold or an adaptive threshold module that dynamically adjusts based on historical data. The relative coefficient deviation is calculated internally by two constraint coefficients within the current control unit. It can be used to quantify the degree of difference between the calculated dynamic constraint coefficient and the estimated constraint coefficient, and to evaluate the consistency between model predictions and measured characteristics. It is understood that existing solutions do not perform consistency verification between the theoretical model output and the actual data-driven results, which may lead to erroneous fusion or runaway response. In this embodiment, the relative coefficient deviation can be obtained by receiving the calculated dynamic constraint coefficient and the estimated constraint coefficient, and calculating the ratio deviation in the form of an absolute difference or a relative percentage between the two to obtain a normalized or absolute deviation value. For example, the relative coefficient deviation may include, but is not limited to, one or more of the following: a ratio deviation that can be defined as the absolute difference between the two or expressed as a relative percentage.

[0129] A preset relative deviation threshold is input to the current control unit and participates in the conditional judgment logic. This threshold can be used to determine whether the difference between the calculated dynamic constraint coefficient and the estimated constraint coefficient is within an acceptable range, thus deciding whether to adopt a weighted fusion strategy. It is understood that traditional methods directly perform fixed-weight fusion, lacking the ability to dynamically adjust the fusion strategy based on data consistency. In a specific embodiment, the preset relative deviation threshold can be a threshold parameter received from system calibration or online learning. By comparing it with the relative coefficient deviation, a decision signal is output to trigger either a weighted average or a single coefficient selection.

[0130] The preset relative deviation threshold can be, but is not limited to, one or more of the following: a static constant or a dynamic threshold generated through self-learning of operating conditions. The weighted average value is calculated and generated internally by the current control unit. As a candidate value for the target limiting coefficient, it can be used to generate an intermediate limiting coefficient that balances theoretical and measured information when there is a significant deviation between the calculated dynamic limiting coefficient and the estimated limiting coefficient. It is understood that traditional weighted fusion uses fixed weights, which cannot adjust the contribution ratio of the model and data according to changes in operating conditions, affecting control accuracy. In this embodiment, the weighted average value can be obtained by receiving the calculated dynamic limiting coefficient, the estimated limiting coefficient, and their corresponding weight coefficients, and fusing the two information through a linear combination to obtain the fused intermediate limiting coefficient. For example, the weighted average value can be, but is not limited to, one or more of the following: determining the weight allocation based on load type confidence, or introducing a fuzzy logic controller to dynamically adjust the weights.

[0131] For example, in a scenario where instantaneous fluctuations in the power grid cause sensor signal distortion during the start-up of a wind turbine motor, the motor starting current limiting circuit based on the soft starter cabinet in this embodiment can prevent the load type confidence level output by the data processing unit from falling below a preset confidence threshold due to waveform distortion. Based on this, the current control unit skips the data-driven path and directly selects the calculated dynamic limiting coefficient output by the parameter calculation unit as the target limiting coefficient, avoiding the misuse of unreliable estimated limiting coefficients. When the power grid stabilizes and the load type confidence level rises and exceeds the threshold, the system re-activates the feature classification results and compares the relative coefficient deviation between the calculated dynamic limiting coefficient and the updated estimated limiting coefficient. If the deviation is small, it switches to the estimated limiting coefficient to accelerate the response; if the deviation is large, it uses a weighted average to achieve a smooth transition, ensuring control continuity and safety.

[0132] This embodiment determines the target limiting coefficient based on load type confidence, calculated dynamic limiting coefficient, and estimated limiting coefficient. Specifically, it uses the calculated dynamic limiting coefficient when the load type confidence is below a preset threshold, and determines whether to use the estimated limiting coefficient or a weighted average when the confidence is above or equal to the threshold, based on the magnitude of the relative coefficient deviation. A preset confidence threshold serves as the decision boundary to trigger the selection of different control paths. The relative coefficient deviation reflects the consistency level between the model output and the measured data. A preset relative deviation threshold determines whether to enable the weighted fusion mechanism. The weighted average generates an intermediate coefficient that balances theoretical rigor and data adaptability when the deviation is large. This achieves the technical effect of prioritizing control stability when load identification uncertainty is high, improving response speed when identification is reliable and the model and data are consistent, and achieving smooth fusion when there are significant differences between the two. Therefore, it effectively improves current limiting accuracy under complex operating conditions, reduces the probability of false overcurrent protection triggering, enhances system robustness and adaptability, and ensures the smoothness and reliability of high-power motor startup.

[0133] In one embodiment, the current limiting current is determined based on the target limiting factor, the motor rated current, and the thyristor parameters of the soft starter cabinet, including:

[0134] The target coefficient current curve is determined from the preset coefficient current relationship library based on the rated current of the motor and the rated firing angle parameters of the thyristors in the soft starter cabinet.

[0135] The retrieved current value corresponding to the target limiting coefficient is determined based on the target coefficient current curve, and the retrieved current value is determined as the current limiting current.

[0136] The coefficient-current relationship library, integrated within the current control unit or connected to it via electrical signals, stores the nonlinear mapping relationship between the target limiting coefficient and the corresponding limiting current under different combinations of motor rated current and thyristor rated firing angle parameters, providing a basis for accurate current value retrieval. It is understandable that traditional current limiting control uses linear proportional conversion or fixed lookup table interpolation methods, which struggle to accurately reflect the nonlinear relationship between the thyristor conduction characteristics and the motor load, leading to decreased current limiting accuracy.

[0137] In one embodiment, the coefficient-current relationship library can receive the rated current of the motor and the rated firing angle parameters of the thyristors in the soft starter cabinet. It generates a current-coefficient relationship curve adapted to the current configuration by indexing the rated current and firing angle dimensions through a multidimensional lookup table built based on historical experimental data or offline simulation, thus obtaining the target coefficient-current curve. The coefficient-current relationship library can, but is not limited to, storing the relationship curve using a piecewise polynomial fitting data structure, or organizing one or more coefficient-current mapping matrices under multiple operating conditions in the form of a three-dimensional array. The target coefficient-current curve, generated by the coefficient-current relationship library and transmitted to the retrieval module in the current control unit, can be used to represent the specific functional relationship between the target limiting coefficient and the limiting current under specific motor rated current and thyristor rated firing angle conditions, for accurately determining the current limiting current.

[0138] It is understandable that directly using a uniform scaling factor to convert the limiting coefficient into a current value fails to consider the differentiated impact of different power levels and thyristor operating points on output characteristics, resulting in control deviations. In this embodiment, the target coefficient current curve can be a query result from a coefficient current relationship database. Through this continuous or discrete function curve, a matching operation is performed after the target limiting coefficient is calculated in real time, outputting a functional relationship expression for retrieving current values.

[0139] In scenarios where high-power wind turbine motors are started under different thyristor configurations, the motor starting current limiting circuit based on the soft starter cabinet in this embodiment can, when the system detects that the current rated current of the motor is large and the thyristor has a small rated firing angle, retrieve the corresponding high-resolution target coefficient current curve based on these two parameters from the coefficient current relationship library; the target limiting coefficient calculated in real time is input to the curve for precise matching to obtain the corresponding retrieved current value; since the curve has taken into account the nonlinear conduction characteristics of the device, the determined current limiting current can more accurately control the slope of the thyristor output voltage, avoid current jumps in the initial stage of startup or insufficient rise in the later stage, and achieve a smooth and reliable soft start process.

[0140] This embodiment determines the target coefficient current curve from a preset coefficient current relationship library based on the motor's rated current and the rated firing angle parameters of the thyristors in the soft starter cabinet. It then determines the retrieved current value corresponding to the target limiting coefficient as the current limiting current based on the target coefficient current curve. The coefficient current relationship library stores the nonlinear mapping relationship between the target limiting coefficient and the corresponding limiting current under different operating conditions. Combined with the motor's rated current and the thyristor's rated firing angle parameters, a target coefficient current curve adapted to the hardware characteristics is dynamically generated, ensuring that the current limiting strategy matches the actual operating characteristics of the main circuit devices. Furthermore, by accurately retrieving the corresponding retrieved current value based on the target limiting coefficient on this curve, a high-fidelity conversion from abstract control parameters to specific physical current values ​​is achieved. This avoids the nonlinear errors caused by traditional linear proportional or table lookup-based coarse interpolation, improving the accuracy and response consistency of current limiting commands. Especially during the startup of high-power motors, it can accurately coordinate the thyristor conduction behavior with load demands, effectively suppressing current surges and preventing overcurrent protection malfunctions or thyristor thermal stress accumulation caused by improper current limiting settings. This significantly enhances the system's control robustness and equipment operational reliability under complex dynamic conditions.

[0141] In one embodiment, limiting the output current of the soft starter cabinet based on the current limiting current includes:

[0142] In response to the soft starter cabinet's output current request value being less than or equal to the current limit current, the soft starter cabinet is controlled to adjust the thyristor firing angle according to the output current request value;

[0143] In response to the soft starter cabinet's output current request value exceeding the current limit, the soft starter cabinet is controlled to adjust the thyristor firing angle according to the current limit and issue an overcurrent warning signal.

[0144] The output current request value, input to the current control unit and participating in the trigger angle adjustment logic, can represent the target current value that the soft starter cabinet expects to output based on the start-up curve or upper-level instructions. This target current is then compared with the current limiting current to determine the control mode. Alternatively, it can receive input from a preset start-up program curve, a remote control system, or a human-machine interface, serving as the criterion for determining whether to perform current limiting intervention.

[0145] In one embodiment, the thyristor firing angle is calculated by the current control unit and sent to the thyristor drive circuit. The drive circuit controls the thyristor turn-on timing, which can be used to control the output voltage and current amplitude by adjusting the thyristor conduction time, thereby achieving continuous regulation of the soft starter cabinet's output current. Alternatively, it can receive control commands (corresponding to the target current) output by the current control unit, and issue a trigger pulse after a certain electrical angle delay based on the AC power supply phase synchronization signal, changing the conduction interval of each cycle to adjust the output power and obtain the actual conduction start point of the thyristor in each cycle. For example, the thyristor firing angle can be achieved using one or more methods, such as zero-crossing synchronous timing combined with phase delay control, or high-precision triggering based on a digital phase-locked loop.

[0146] Furthermore, the overcurrent warning signal is generated by the current control unit and output to the upper-level monitoring system, alarm device, or recording module. It can be used to generate a warning signal when the requested output current value exceeds the current limit, indicating a potential overcurrent risk in the system. It can receive the comparison result of the requested output current value being greater than the current limit and output a warning indication in the form of a digital switch signal or a communication message. It is understood that traditional soft starters typically only trigger protection shutdown after the actual current exceeds the limit, lacking a proactive warning mechanism and making it difficult to support preventative maintenance. In this embodiment, the overcurrent warning signal can be, but is not limited to, configured as a relay contact output, or sent via a communication interface as one or more MODBUS alarm codes.

[0147] For example, in a scenario where a sudden increase in load occurs during the startup of a wind turbine motor, the motor starting current limiting circuit based on the soft starter cabinet in this embodiment can generate an output current request value according to a preset curve. When the damper is accidentally closed, causing the load to rise instantaneously, this request value will rise rapidly. When the request value exceeds the current limiting current determined by multi-source fusion, the current control unit no longer adjusts the thyristor firing angle according to the request value, but recalculates and locks the maximum allowable conduction angle based on the current limiting current to suppress further increase in output current. At the same time, an overcurrent warning signal is issued to notify the control system to check or adjust the operating conditions to avoid thyristor thermal damage or tripping accidents caused by continuous overcurrent.

[0148] This embodiment controls the soft starter to adjust the thyristor firing angle according to the output current request value when it is less than or equal to the current limit current, and controls the soft starter to adjust the thyristor firing angle according to the current limit current and issue an overcurrent warning signal when the output current request value is greater than the current limit current. The output current request value is used as a judgment benchmark and input to the current control unit to participate in the control logic operation. The thyristor firing angle is adjusted based on phase synchronization and delay control to adjust the conduction interval to achieve continuous regulation of the output current. The overcurrent warning signal is generated and uploaded to the monitoring system when the request value exceeds the limit to provide early risk warning. It can achieve the technical effects of fine current control based on the dynamic relationship between the output current demand and the limit threshold, actively suppress current growth under transient overcurrent conditions and provide warning information, enhance the robustness of the system under abnormal operating conditions, reduce unplanned downtime and improve equipment safety and operational continuity.

[0149] Furthermore, to achieve the above objectives, the present invention also proposes a motor starting current limiting device based on a soft starter cabinet, wherein the motor starting current limiting device based on a soft starter cabinet includes the motor starting current limiting circuit based on a soft starter cabinet described above. The specific structure of this motor starting current limiting circuit based on a soft starter cabinet is as described in the above embodiments. Since this motor starting current limiting device based on a soft starter cabinet adopts all the technical solutions of all the above embodiments, it possesses at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be elaborated upon here.

[0150] Furthermore, to achieve the above objectives, the present invention also proposes a soft starter cabinet, which includes the motor starting current limiting circuit based on the soft starter cabinet described above. The specific structure of this motor starting current limiting circuit based on the soft starter cabinet is as described in the above embodiments. Since this soft starter cabinet adopts all the technical solutions of all the above embodiments, it possesses at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be elaborated upon here.

[0151] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A motor starting current limiting circuit based on a soft starter cabinet, characterized in that, The circuit includes: a parameter calculation unit, a data processing unit, and a current control unit; The parameter calculation unit is connected to the data processing unit and the current control unit, respectively, and the current control unit is connected to the data processing unit and the soft starter cabinet, respectively. The parameter calculation unit is used to determine the reference current based on the motor rated current, the grid voltage fluctuation value and the load type parameter, determine the current overcurrent rate based on the reference current and the actual starting current, and determine the calculated dynamic limit coefficient based on the current overcurrent rate and the preset motor dynamic model. The calculated dynamic limit coefficient is a calculated dynamic limit coefficient that reflects the change characteristics of motor impedance and back electromotive force, derived based on the current overcurrent rate and the motor dynamic model. The data processing unit is used to correct the sampling area of ​​the current sensor detection data according to the reference current and the soft starter cabinet response delay to obtain valid detection data, and to perform feature classification on the valid detection data to obtain the estimated limitation coefficient and load type confidence. The current control unit is used to determine the target limiting coefficient based on the load type confidence level, the calculated dynamic limiting coefficient, and the estimated limiting coefficient, and to determine the current limiting current based on the target limiting coefficient, the motor rated current, and the thyristor parameters of the soft starter cabinet, and to limit and control the output current of the soft starter cabinet based on the current limiting current.

2. The motor starting current limiting circuit based on a soft starter cabinet as described in claim 1, characterized in that, The determination of the reference current based on the motor's rated current, grid voltage fluctuation value, and load type parameters includes: The average reference current is determined based on the motor's rated current and the grid voltage fluctuation value, and the voltage fluctuation coefficient is determined based on the average reference current and the grid voltage fluctuation value. The load correction factor is determined based on the load type parameters and the preset load type library; The reference current is determined from the average reference current, the motor rated current, and the load correction current based on the voltage fluctuation coefficient and the load correction coefficient.

3. The motor starting current limiting circuit based on a soft starter cabinet as described in claim 2, characterized in that, The step of determining the reference current from the average reference current, the motor rated current, and the load correction current based on the voltage fluctuation coefficient and the load correction coefficient includes: In response to the voltage fluctuation coefficient being less than a preset first fluctuation threshold, the average reference current is determined as the reference current; In response to the load correction factor being greater than or equal to a preset load factor threshold and the voltage fluctuation factor being greater than or equal to a preset first fluctuation threshold, a correction value of 1.2 times the rated current of the motor is determined as the reference current. In response to the load correction coefficient being less than a preset load coefficient threshold, and the voltage fluctuation coefficient being greater than or equal to a preset first fluctuation threshold and less than a preset second fluctuation threshold, the weighted average of the average reference current and the load correction current is determined as the reference current. In response to the load correction coefficient being less than a preset load coefficient threshold and the voltage fluctuation coefficient being greater than or equal to a preset second fluctuation threshold, a voltage correction current corresponding to the rated current of the motor is determined according to a preset voltage and current compensation model, and the voltage correction current is determined as the reference current.

4. The motor starting current limiting circuit based on a soft starter cabinet as described in claim 1, characterized in that, The step of correcting the sampling area of ​​the current sensor detection data based on the reference current and the soft starter cabinet response delay to obtain valid detection data includes: Determine the trigger delay time of the current soft starter cabinet and the sampling window parameters of the sensor detection data; determine the current sampling delay compensation amount based on the reference current and the trigger delay time; The current sampling window offset is determined based on the sampling window parameters and the load type parameters; The sampling region is corrected on the time axis according to the sampling delay compensation amount and the sampling window offset to obtain the corrected sampling window, and the current sensor detection data is truncated according to the corrected sampling window to obtain the effective detection data.

5. The motor starting current limiting circuit based on a soft starter cabinet as described in claim 1, characterized in that, The step of performing feature classification on the effective detection data to obtain the estimated constraint coefficient and load type confidence score includes: Feature extraction is performed on the effective detection data to obtain the current rise slope, steady-state current value, and temperature change rate; The current rise slope, steady-state current value, and temperature change rate are input into a preset load classification model to obtain the load type and the load type confidence level corresponding to the effective detection data. The retrieval coefficient corresponding to the load type is determined according to the preset type coefficient mapping relationship, and the retrieval coefficient is determined as the estimated constraint coefficient.

6. The motor starting current limiting circuit based on a soft starter cabinet as described in claim 1, characterized in that, The step of determining the target limiting coefficient based on the load type confidence level, the calculated dynamic limiting coefficient, and the estimated limiting coefficient includes: In response to the load type confidence level being less than a preset confidence threshold, the calculated dynamic constraint coefficient is determined as the target constraint coefficient; In response to the load type confidence level being greater than or equal to a preset confidence threshold, the relative coefficient deviation is determined based on the calculated dynamic constraint coefficient and the estimated constraint coefficient; In response to the relative coefficient deviation being less than a preset relative deviation threshold, the estimated constraint coefficient is determined as the target constraint coefficient; In response to the relative coefficient deviation being greater than or equal to a preset relative deviation threshold, the weighted average of the calculated dynamic constraint coefficient and the estimated constraint coefficient is determined as the target constraint coefficient.

7. The motor starting current limiting circuit based on a soft starter cabinet as described in claim 1, characterized in that, The step of determining the current limiting current based on the target limiting coefficient, the motor rated current, and the thyristor parameters of the soft starter cabinet includes: The target coefficient current curve is determined from the preset coefficient current relationship library based on the rated current of the motor and the rated firing angle parameters of the thyristor of the soft starter cabinet. The retrieved current value corresponding to the target limiting coefficient is determined based on the target coefficient current curve, and the retrieved current value is determined as the current limiting current.

8. The motor starting current limiting circuit based on a soft starter cabinet as described in claim 1, characterized in that, The step of limiting and controlling the output current of the soft starter cabinet based on the current limiting current includes: In response to the soft starter cabinet's output current request value being less than or equal to the current limit current, the soft starter cabinet is controlled to adjust the thyristor firing angle according to the output current request value; In response to the soft starter cabinet's output current request value being greater than the current limit current, the soft starter cabinet is controlled to adjust the thyristor firing angle according to the current limit current and issue an overcurrent warning signal.

9. A motor starting current limiting device based on a soft starter cabinet, characterized in that, The motor starting current limiting device based on the soft starter cabinet includes the motor starting current limiting circuit based on the soft starter cabinet as described in any one of claims 1 to 8.

10. A soft starter cabinet, characterized in that, The soft starter cabinet includes the motor starting current limiting circuit based on the soft starter cabinet as described in any one of claims 1 to 8.

Images