Load power track simulation control method and device and medium

Abstract

The invention relates to the technical field of power supply equipment power control. The load power track simulation control method comprises the steps of obtaining voltage information and current information of target power supply equipment, determining power deviation information and a dynamic target voltage range generated on the basis of a preset load behavior prediction model, constructing a combined regulation and control weight vector on the basis of the power deviation information, and obtaining a load power track simulation control result. According to the combined regulation and control weight vector and the dynamic target voltage range, constructing a target power evolution trajectory curve by using a physical response inversion mechanism, executing time discretization processing on the target regulation and control trajectory curve, extracting voltage change values and current change values corresponding to each time step, and calculating the target power evolution trajectory curve; and combining the voltage change value and the current change value according to a time sequence to form a control parameter sequence, and generating an analog control instruction matched with the target power evolution trajectory curve according to the control parameter sequence. The method has the effect of improving the real-time response capability of the power supply equipment under complex working conditions.

Description

Technical Field

[0001] This invention relates to the technical field of power control for power supply equipment, and in particular to a load power trajectory simulation control method, device, and medium. Background Technology

[0002] In the fields of intelligent power management and power electronic control, the simulation and regulation of load power variation trends have become one of the core factors affecting system response performance and operational stability. Especially in complex power grid environments or nonlinear load scenarios, power supply equipment faces frequent dynamic power disturbances and transient voltage fluctuations. Traditional control methods based on linear strategies or static models are no longer sufficient to meet the requirements for refined modeling and real-time response control of load power trajectories.

[0003] Most existing methods rely on fixed-rule voltage-current control logic or experience-based regulation strategies, which cannot effectively reflect the dynamic response of the load to different regulation behaviors. Although some methods using load prediction models can estimate future power trends to a certain extent, they are usually limited to numerical output and lack a control command generation mechanism corresponding to the target power behavior, making it difficult to form a precise closed loop for regulating voltage and current signals. Summary of the Invention

[0004] To improve the real-time response capability of power supply equipment under complex operating conditions, this application provides a load power trajectory simulation control method, device and medium.

[0005] The above-mentioned objective of this application is achieved through the following technical solution: A load power trajectory simulation control method, the load power trajectory simulation control method comprising: Acquire voltage and current information of the target power supply device, determine power deviation information, and generate a dynamic target voltage range based on a preset load behavior prediction model; Based on the power deviation information, a joint control weight vector is constructed; Based on the joint control weight vector and the dynamic target voltage range, the target power evolution trajectory curve is constructed using the physical response inversion mechanism; Perform time discretization processing on the target control trajectory curve to extract the voltage change value and current change value corresponding to each time step; The voltage change value and the current change value are combined in time sequence to form a control parameter sequence. Based on the control parameter sequence, instruction encoding is performed according to preset hardware and software interface constraints to generate an analog control instruction that matches the target power evolution trajectory curve.

[0006] By adopting the above technical solution, power deviation information can be extracted based on the real-time voltage and current information of the target power supply equipment. Combined with the trained load behavior prediction model, a dynamic target voltage range can be generated. Then, a joint control weight vector that fits the electrical characteristics of the load can be constructed. The target power evolution trajectory curve closely related to the voltage and current regulation path can be generated using the physical response inversion mechanism. Through trajectory time discretization and control parameter sequence encoding, the target power trajectory can be accurately mapped into control commands that conform to the hardware and software interface constraints. This enables continuous modeling and synchronous control of the power evolution process in complex power grids or nonlinear load scenarios. It overcomes the bottleneck problems of response lag and disconnect between prediction output and control execution in existing methods, effectively enhancing the real-time performance and dynamic stability of power simulation control.

[0007] In a preferred example, this application can be further configured such that the preset load behavior prediction model previously includes: Acquire historical voltage and current information under multiple typical load scenarios; The historical voltage information and the historical current information are arranged in chronological order to construct an input sequence, and the target power change trend and control response deviation information corresponding to the input sequence are obtained. Based on the input sequence, the target power change trend, and the control response deviation information, a joint training sample set is constructed; the parameters of the model are trained using the joint training sample set; during the training process, the power prediction error information is calculated based on the difference between the predicted power value output by the model and the target power change trend, and the control response deviation information is calculated based on the error between the voltage response curve generated by the model and the target response curve. Based on a preset weighted multi-objective loss function, the power prediction error information and the control response deviation information are jointly optimized to obtain the preset load behavior prediction model.

[0008] By adopting the above technical solution, the historical behavior patterns of voltage and current can be fully extracted under multiple typical load scenarios. By constructing a joint mapping relationship between time series input and target response trend, a dual training structure covering power evolution trend and control response deviation can be effectively established. Furthermore, a weighted multi-objective loss function is used to jointly optimize prediction accuracy and control consistency, ultimately forming a load behavior prediction model that can characterize the dynamic behavior of the load and has the ability to adjust and adapt. This improves the power system's response prediction capability to nonlinear load disturbances and solves the problems of insufficient load modeling accuracy and lack of control deviation feedback in existing solutions.

[0009] In a preferred embodiment, this application can be further configured such that the dynamic target voltage range generated based on a preset load behavior prediction model includes: The voltage information and the current information are constructed into a multi-dimensional input sequence according to a preset time sequence, and then input into the preset load behavior prediction model to obtain the power change trend of the target power supply device within the target prediction period. Based on the power change trend, a prediction offset correction is performed in conjunction with the historical steady-state voltage output range to obtain the expected dynamic output voltage range. The upper and lower boundaries of the dynamic output voltage range are calibrated to generate the dynamic target voltage range that meets the stable response requirements of the current load behavior.

[0010] By adopting the above technical solution, real-time voltage and current information can be constructed into a multi-dimensional input structure with behavioral correlation characteristics while ensuring timing consistency. This enhances the load behavior prediction model's ability to capture dynamic power trends. Furthermore, based on the prediction results, historical steady-state voltage output range information is integrated to perform offset correction operations, effectively correcting the systematic deviation between the model output and the actual response. Finally, through upper and lower boundary calibration operations, the target voltage range can be dynamically adapted while meeting the current load's stable response characteristics, thereby improving the compatibility and accuracy of the control strategy for complex load behaviors.

[0011] In a preferred embodiment, this application can be further configured such that: the construction of the joint regulation weight vector based on the power deviation information includes: Sliding statistical processing is performed on the power deviation information within a preset short time window and a preset long time window respectively to extract short-term fluctuation features and long-term trend features. The voltage regulation priority factor is determined based on the short-term fluctuation characteristics, and the current regulation sensitivity factor is determined based on the long-term trend characteristics. The voltage regulation priority factor and the current regulation sensitivity factor are combined according to a weighted strategy to generate the joint regulation weight vector.

[0012] By adopting the above technical solution, power deviation information can be processed by sliding statistics at multiple time scales. Within a short time window, rapid fluctuation patterns can be effectively captured to extract short-term fluctuation features, and stable trends can be identified within a long time window to extract long-term trend features. Then, based on the short-term fluctuation features, the urgency of voltage regulation response can be evaluated and the voltage regulation priority factor can be determined. Based on the long-term trend features, the persistence of load behavior changes can be judged and the current adjustment sensitivity factor can be determined. By weighted combination of the two types of factors, a joint control weight vector is generated, so that the control strategy can take into account both immediacy and stability in dynamic adjustment.

[0013] In a preferred embodiment, this application can be further configured as follows: the step of constructing the target power evolution trajectory curve using a physical response inversion mechanism based on the joint control weight vector and the dynamic target voltage range includes: Based on the joint control weight vector, the voltage change and current response amplitude corresponding to each stage in the voltage regulation path are calculated, and a voltage change path is generated based on the dynamic target voltage range. Based on the preset load electrical response relationship, a response inversion operation is performed on the voltage change path to obtain the current response path; the voltage change amount and the current response amplitude are multiplied by the control time step to obtain the initial power evolution trajectory; the initial power evolution trajectory is compared with the target power change trend point by point to extract the power deviation value at each control time step. Based on the power deviation value, the corresponding voltage change and current response amplitude are adjusted in reverse, and the voltage change path and current response path are updated synchronously to obtain the updated voltage change path and updated current response path. The power evolution trajectory is recalculated based on the updated voltage change path and the updated current response path. The power evolution trajectory is then repeatedly compared and corrected with the target power change trend until the power deviation in each control time step meets the preset convergence condition. Finally, the target power evolution trajectory curve is determined.

[0014] By adopting the above technical solution, a voltage change path can be constructed based on the joint control weight vector and staged voltage regulation can be performed in combination with the dynamic target voltage range. Then, the corresponding current response path can be deduced through the load electrical response relationship. On this basis, the voltage change and current response amplitude are multiplied point by point according to the control time step to obtain the initial power evolution trajectory. The error is compared with the target power change trend to extract the power deviation value at each moment. The path is corrected by adjusting the voltage change and current response amplitude in a targeted manner to form an updated voltage and current response path. The power evolution trajectory is then reconstructed iteratively by the corrected path and continuously compared until the error meets the convergence condition, thereby dynamically approximating the target power change trend and realizing high-precision simulation modeling and closed-loop control command support for load power behavior.

[0015] In a preferred embodiment, this application can be further configured as follows: the step of generating analog control instructions that match the target power evolution trajectory curve by performing instruction encoding processing according to the control parameter sequence and preset software and hardware interface constraints includes: Based on the hardware and software interface constraints, determine the field position and signal representation format of each control parameter pair in the control parameter sequence within the target control command structure; Based on the signal representation format, each pair of control parameters is converted into a recognizable numerical control signal and embedded into the control instruction structure according to the field position to obtain an initial instruction set containing multi-time control information. The initial instruction set is serialized in chronological order, and a synchronization flag field is inserted according to the preset hardware and software interface constraints to generate the simulation control instructions.

[0016] By adopting the above technical solution, the structural mapping and numerical conversion of the control parameter pairs in the control parameter sequence can be performed in combination with the constraints of the software and hardware interface. The field position and signal representation format of each parameter pair in the target control command structure are clarified, so that the voltage change value and current change value are accurately encoded into numerical control signals and embedded into the control command structure to form a complete initial command set. Further, the serialization process is performed through time sequence, and a synchronization mark field is embedded on the basis of meeting the interface rhythm and synchronization logic requirements. Finally, an analog control command with time consistency and signal recognizability is generated, realizing the efficient mapping of power trajectory to execution-level command and ensuring a fine match between control behavior and target power evolution curve.

[0017] In a preferred embodiment, this application can be further configured as follows: the initial instruction set is serialized in chronological order, and a synchronization flag field is inserted according to the preset hardware and software interface constraints to generate the simulation control instructions, including: The time tags of each control parameter pair in the initial instruction set are sorted in ascending order to obtain a time-sequential instruction stream; the insertion strategy of the synchronization mark field in the instruction stream is determined according to the preset hardware and software interface constraints, and an insertion position list is generated. In the insertion position list, a synchronization marker field containing a rhythm identification identifier and a control synchronization flag is embedded. The field structure and encoding method of the synchronization marker field are constrained and configured to form the analog control command.

[0018] By adopting the above technical solution, an ascending sorting operation can be performed based on the time tags of each control parameter pair in the initial instruction set to construct a time-seriesd instruction stream with a clear execution order. Based on this, combined with preset hardware and software interface constraints, the insertion strategy of the synchronization mark field is determined, thereby generating an insertion position list. The synchronization mark field containing rhythm identification and control synchronization flags is embedded at key positions in the instruction stream, and a unified standard is applied to the structure and encoding method of the field. This ensures that the final analog control instruction not only has clear time rhythm control capabilities, but also ensures data synchronization and rhythm matching requirements with the underlying actuator or interface device, thereby improving the real-time control accuracy and communicability of the analog instruction at the execution layer.

[0019] The second objective of this invention is achieved through the following technical solution: A load power trajectory simulation control device, the load power trajectory simulation control device comprising: The electrical parameter acquisition and power deviation calculation module is used to acquire voltage and current information of the target power supply equipment, determine power deviation information, and generate a dynamic target voltage range based on a preset load behavior prediction model. A dynamic target voltage generation module is used to construct a joint control weight vector based on the power deviation information; The joint control weight construction module is used to construct the target power evolution trajectory curve based on the joint control weight vector and the dynamic target voltage range using a physical response inversion mechanism. The physical response-driven trajectory reconstruction module is used to perform time discretization processing on the target control trajectory curve and extract the voltage change value and current change value corresponding to each time step. The analog control command generation module is used to combine the voltage change value and the current change value in time sequence to form a control parameter sequence, and perform command encoding processing according to the control parameter sequence and preset software and hardware interface constraints to generate an analog control command that matches the target power evolution trajectory curve.

[0020] By adopting the above technical solution, power deviation information can be extracted based on the real-time voltage and current information of the target power supply equipment. Combined with the trained load behavior prediction model, a dynamic target voltage range can be generated. Then, a joint control weight vector that fits the electrical characteristics of the load can be constructed. The target power evolution trajectory curve closely related to the voltage and current regulation path can be generated using the physical response inversion mechanism. Through trajectory time discretization and control parameter sequence encoding, the target power trajectory can be accurately mapped into control commands that conform to the hardware and software interface constraints. This enables continuous modeling and synchronous control of the power evolution process in complex power grids or nonlinear load scenarios. It overcomes the bottleneck problems of response lag and disconnect between prediction output and control execution in existing methods, improves control granularity and execution consistency, and effectively enhances the real-time performance and dynamic stability of power simulation control.

[0021] The above-mentioned objective three of this application is achieved through the following technical solution: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described load power trajectory simulation control method.

[0022] The fourth objective of this application is achieved through the following technical solution: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described load power trajectory simulation control method.

[0023] In summary, this application includes at least one of the following beneficial technical effects: 1. It can extract power deviation information based on the real-time voltage and current information of the target power supply equipment, and generate a dynamic target voltage range by combining the trained load behavior prediction model. Then, it constructs a joint control weight vector that fits the electrical characteristics of the load, uses the physical response inversion mechanism to generate the target power evolution trajectory curve closely related to the voltage and current regulation path, and accurately maps the target power trajectory into control commands that conform to the hardware and software interface constraints through trajectory time discretization and control parameter sequence encoding. This enables continuous modeling and synchronous control of the power evolution process in complex power grids or nonlinear load scenarios, overcomes the bottleneck problems of response lag and disconnect between prediction output and control execution in existing methods, improves control granularity and execution consistency, and effectively enhances the real-time performance and dynamic stability of power simulation control. 2. It can perform structural mapping and numerical conversion of the control parameter pairs in the control parameter sequence by combining hardware and software interface constraints, clarify the field position and signal representation format of each parameter pair in the target control command structure, so as to accurately encode the voltage change value and current change value into numerical control signal and embed it into the control command structure to form a complete initial command set; further, it performs serialization processing through time sequence, and embeds a synchronization mark field on the basis of meeting the interface rhythm and synchronization logic requirements, and finally generates analog control commands with time consistency and signal recognizability, realizing efficient mapping of power trajectory to execution level commands, and ensuring fine matching between control behavior and target power evolution curve; 3. It can perform ascending sorting operations based on the time tags of each control parameter pair in the initial instruction set, construct a time-seriesd instruction stream with a clear execution order, and on this basis, combined with preset hardware and software interface constraints, determine the insertion strategy of the synchronization mark field, thereby generating an insertion position list. The synchronization mark field containing rhythm identification and control synchronization flags is embedded at key positions in the instruction stream, and a unified standard is applied to the structure and encoding method of the field. This ensures that the final analog control instruction not only has clear time rhythm control capabilities, but also ensures data synchronization and rhythm matching requirements with the underlying actuator or interface device, thereby improving the real-time control accuracy and communicability of the analog instruction at the execution layer. Attached Figure Description

[0024] Figure 1 This is a flowchart of a load power trajectory simulation control method in one embodiment of the application; Figure 2 This is a flowchart illustrating the implementation of step S10 in a load power trajectory simulation control method according to an embodiment of this application. Figure 3 This is another implementation flowchart of step S10 in a load power trajectory simulation control method according to an embodiment of this application; Figure 4 This is a flowchart illustrating the implementation of step S20 in a load power trajectory simulation control method according to an embodiment of this application. Figure 5 This is a flowchart illustrating the implementation of step S30 in a load power trajectory simulation control method according to an embodiment of this application. Figure 6 This is a flowchart illustrating the implementation of step S50 in a load power trajectory simulation control method according to an embodiment of this application. Figure 7 This is a flowchart illustrating the implementation of step S503 in a load power trajectory simulation control method according to an embodiment of this application. Figure 8 This is a schematic block diagram of a load power trajectory simulation control device according to an embodiment of this application; Figure 9 This is a schematic diagram of a device according to one embodiment of this application. Detailed Implementation

[0025] The present application will be further described in detail below with reference to the accompanying drawings.

[0026] In one embodiment, such as Figure 1 As shown, this application discloses a load power trajectory simulation control method, which specifically includes the following steps: S10: Obtain the voltage and current information of the target power supply device, determine the power deviation information, and the dynamic target voltage range generated based on the preset load behavior prediction model.

[0027] In this embodiment, the preset load behavior prediction model refers to a deep time series modeling network based on historical voltage and current sequences and corresponding power evolution trends. This network is used to capture the dynamic mapping relationship between voltage-current input and power response under different load scenarios, and supports trend prediction and response deviation estimation of the power behavior of the target power supply device in the future time window.

[0028] Specifically, the process of acquiring voltage and current information of the target power supply equipment includes: during the operation of the target power supply equipment, acquiring instantaneous voltage values ​​in each control cycle through a voltage acquisition circuit, and synchronously acquiring instantaneous current values ​​in each control cycle through a current sensor, forming a voltage sequence and a current sequence arranged in chronological order. Based on the voltage and current sampling point pairs in the current control cycle, the current power value is calculated using the active power calculation formula, P = U × I × cos(φ), where U is the voltage sampling value, I is the current sampling value, and cos(φ) is the current power factor. The power factor is obtained by synchronously detecting the phase difference between the voltage signal and the current signal. Furthermore, using this current power value and the target power supply equipment from the previous moment... The difference between power values ​​is calculated to obtain power deviation information, which indicates the degree of deviation between the actual power output at the current moment and the target power trajectory. Furthermore, the voltage and current sequences arranged in chronological order are constructed into a multi-dimensional input sequence and input into a preset load behavior prediction model. The load behavior prediction model is a deep network structure with time-series prediction capabilities after training. The input end accepts the multi-dimensional input sequence, and the output end generates a predicted power sequence within a future time window. Then, the predicted power sequence is mapped and associated with the historical steady-state voltage range. The corresponding dynamic target voltage range is determined by a lookup table method or a fitting function. The dynamic target voltage range includes the acceptable upper and lower boundaries of the voltage at each moment within the prediction time window.

[0029] S20: Construct a joint control weight vector based on power deviation information.

[0030] Specifically, multi-scale analysis is performed on the power deviation information acquired within the target control cycle. In a shorter time window, the instantaneous fluctuation characteristics are extracted using a moving average method, where the short-term window length is set to several milliseconds based on the load response frequency characteristics. In the moving average method, the power deviation samples within each window are weighted and the positions of extreme points are recorded. In a longer time window, the trend changes of the power deviation information are analyzed using local linear regression, where the long-term window length is set to seconds. The change curve formed by the power deviation samples is fitted using the least squares method, and the slope is extracted as a trend feature. The priority coefficient for voltage adjustment is determined based on the short-term fluctuation characteristics. This priority coefficient reflects the ability of voltage changes to compensate for instantaneous fluctuations. The response sensitivity coefficient for current adjustment is determined based on the long-term trend characteristics. This response sensitivity coefficient characterizes the ability of current changes to continuously compensate for trend changes. The two coefficients are fused using a linear weighted combination method to form a joint control weight vector.

[0031] S30: Based on the joint control weight vector and the dynamic target voltage range, construct the target power evolution trajectory curve using the physical response inversion mechanism.

[0032] In this embodiment, the physical response inversion mechanism refers to establishing a dynamic mapping relationship between voltage, current and power by setting an initial voltage regulation path and gradually inverting the current response path in each control time step based on the nonlinear coupling relationship between voltage change and current response, thereby constructing a power trajectory that reflects the changing trend of the control target.

[0033] Specifically, firstly, the weight factors in the joint control weight vector are normalized, and the normalization result is used as the relative contribution ratio of voltage adjustment and current response. Then, a set of voltage change paths that meet the adjustment amplitude limit is generated based on the dynamic target voltage range. This path distributes the voltage change to each time step throughout the entire control cycle using time as an index. The voltage change in each time step is adjusted in amplitude using a weight adjustment coefficient. Next, based on the voltage change path, and according to the coupling relationship between the voltage change and the current response amplitude, the current response path is derived by means of a numerical mapping table or function fitting model. The current response amplitude is scaled proportionally according to the current factor in the joint control weight vector. After obtaining the voltage change path and the current response path... The initial power evolution trajectory is constructed by multiplying the voltage change and current response amplitude at each time step. Then, the power evolution trajectory is compared with the target power change trend at each time step. The power deviation between the two is calculated point by point and the error value is recorded. The amplitude coefficients of the aforementioned voltage change and current response amplitude are adjusted according to the error value, and the voltage change path and current response path are regenerated. The process of multiplying to construct the power trajectory and comparing the deviation is repeated. The above path adjustment and deviation evaluation process is continuously iterated until the power deviation at all time steps is lower than the set convergence threshold. Finally, the converged voltage change path and current response path are used as the basis for power trajectory generation, and the target power evolution trajectory curve is reconstructed by multiplying.

[0034] S40: Perform time discretization processing on the target control trajectory curve to extract the voltage and current change values ​​corresponding to each time step.

[0035] Specifically, the target power evolution trajectory curve is viewed as a function representing a continuous change in power over time. After constructing the power trajectory, a set of equally spaced time step division criteria is set. Based on these criteria, the target power trajectory curve is discretized over the entire control cycle. By analyzing the current power value of the power trajectory curve at each time step, and combining the voltage change path function and the current response path function, the corresponding voltage change value and current change value are calculated at the same time point. The voltage change path function and the current response path function can be obtained through the aforementioned response inversion steps. The calculated parameters are obtained and have a clear time input and output response relationship. The calculation process is executed sequentially according to the time index, and finally an ordered sequence of control parameters consisting of voltage change values ​​and current change values ​​is formed in the entire control cycle. For example, if the total duration of the target power trajectory is 100 milliseconds and the control interval is set to 5 milliseconds, it is divided into 20 time steps. At the 10th time point, i.e., 50 milliseconds, the calculated voltage change value is 2.3 volts and the current change value is 1.1 amps. The same calculation operation is continued to be performed on the remaining time steps to obtain the voltage and current control parameters corresponding to each moment in the entire control process.

[0036] S50: Combines voltage and current changes in chronological order to form a control parameter sequence. Based on the control parameter sequence, performs instruction encoding processing according to preset hardware and software interface constraints to generate analog control instructions that match the target power evolution trajectory curve.

[0037] In this embodiment, the preset hardware and software interface constraints refer to the structural specifications and signal representation restrictions for the generation of control commands, which are uniformly determined according to the input protocol standard, voltage and current analog output capability, level encoding method, instruction structure format and communication rhythm of the target control platform. These constraints include the specific position of each control parameter field in the instruction structure, the encoding word length, the signal amplitude mapping rule, the position and quantity limit of the time synchronization mark, etc., to ensure that the generated analog control commands can be correctly parsed by the control interface and drive the power supply behavior to execute.

[0038] Specifically, voltage and current changes are paired sequentially according to time, forming a continuous sequence of control parameter pairs. Based on this sequence, the field order, byte offset, and required bit width format of each control parameter pair in the target control instruction structure are first determined according to preset hardware and software interface constraints. Then, a signal format conversion operation is performed for each control parameter. For example, the floating-point voltage change value is multiplied by 1024 and rounded to form a 12-bit unsigned integer using fixed-point encoding, or the current change value is mapped to an 8-bit control signal value through linear normalization. After format conversion, the converted signal is embedded into the byte structure corresponding to each control instruction according to the field offset order, forming an initial instruction set indexed by time step. The initial instruction set is then sorted according to the control time sequence and combined with the hardware and software interface protocol. The timing synchronization field insertion method defined in the protocol inserts specific synchronization marker fields at preset rhythm intervals, ultimately constructing a complete and timing-consistent set of control instructions to drive power supply equipment to perform analog output behavior consistent with the target power evolution trajectory. For example, for the third time step in the control parameter sequence, the voltage change value is +0.15V and the current change value is -0.3A. Then, +0.15V is encoded as "000011001010" according to the fixed-point scaling rule, and -0.3A is normalized to "11101010". In the control instruction structure, the voltage field is located in the first 6 bytes and the current field is located in the 7th byte. The above two values ​​are embedded into the corresponding field positions to form the control instruction byte stream "0CA8E3…" corresponding to the third time step. Then, "AA55" is inserted as a synchronization marker to achieve precise alignment of control behavior.

[0039] In one embodiment, such as Figure 2 As shown, in step S10, i.e., the preset load behavior prediction model, the following steps are included: S101: Obtain historical voltage and current information under multiple typical load scenarios.

[0040] Specifically, by backtracking historical data of typical electrical equipment under different operating conditions, scenarios including no-load operation, stable load operation, load change response, and periodic power fluctuation are selected. In each scenario, the voltage change curve and current change curve of the target power supply equipment output terminal are recorded. Data can be acquired by connecting the acquisition device to the voltage sampling node and the current sampling node, and the voltage and current values ​​are collected at equal time intervals in each sampling period.

[0041] S102: Construct an input sequence from historical voltage and current information in chronological order, and obtain the target power change trend and control response deviation information corresponding to the input sequence.

[0042] Specifically, based on the collected historical voltage and current information, the voltage and current values ​​at each time point should be paired and combined sequentially according to the same time sampling period to construct a time series structure reflecting the change law of electrical state. The data pair at each time point includes the voltage and current values ​​corresponding to that moment. By arranging the voltage and current pairs at multiple consecutive time points in chronological order, a complete input sequence can be formed. After constructing the input sequence, the instantaneous power can be calculated based on the voltage and current values ​​at each time point, and the target power change trend can be formed based on the continuous change of this power over time. After obtaining the target power change trend, the response error value at each time point is calculated by comparing the voltage response data output by the actual device under the same input conditions with the preset reference voltage response data. This results in a control response deviation information sequence covering the entire input sequence. This control response deviation information can be used to measure the consistency between the model output behavior and the target behavior during subsequent training. S103: Construct a joint training sample set based on the input sequence, the target power change trend, and the control response deviation information.

[0043] Specifically, the voltage and current pairing data in each consecutive time period of the input sequence should be used as feature vectors, and index matching processing should be performed with the target power change trend curve and control response deviation signal in the same time period. In the matching process, the electrical parameter data in the input sequence and the target power change trend sequence are first time-aligned according to the time label to ensure that each input feature vector corresponds to a unique target power value and response deviation value. Then, multiple sets of three-element sample units are constructed by combination. That is, each sample unit contains a set of input electrical feature vectors, a target power value at a corresponding time, and a control response deviation value. All sample units are arranged in time to form a complete joint training sample set.

[0044] S104: The parameters of the model are trained using the joint training sample set. During the training process, the power prediction error information is calculated based on the difference between the predicted power value output by the model and the target power change trend. The control response deviation information is calculated based on the error between the voltage response curve generated by the model and the target response curve.

[0045] Specifically, the input feature vectors in the joint training sample set should be used as the input data of the model. After initializing the model structure and setting hyperparameters such as the number of training rounds and the learning rate, a forward propagation operation is performed. In the forward propagation, the model performs a nonlinear mapping on the input feature vectors based on the current parameter weights, outputting the predicted power value and the corresponding voltage response curve. Then, the time-by-time difference between the predicted power value and the target power change trend marked in the joint training sample set is calculated to obtain the power prediction error at each time step. The voltage response curve generated by the model is then aligned and compared with the corresponding target response curve in the sample set. The response error of each segment is extracted according to the time step to form the control response deviation information. In each training round, the power prediction error and the control response deviation information are used as sub-terms of the loss function, respectively, and the weighted joint loss value is calculated. The model weights are then updated based on the backpropagation mechanism. The above training process is repeated until the convergence condition is met or the preset number of training rounds is reached, and finally the optimized model parameters are obtained.

[0046] S105: Based on a preset weighted multi-objective loss function, the power prediction error information and control response deviation information are jointly optimized to obtain a preset load behavior prediction model.

[0047] In this embodiment, the preset weighted multi-objective loss function refers to a composite loss function structure that simultaneously measures power prediction error and control response deviation. It includes two sub-loss functions: a power prediction error term and a control response deviation term. The power prediction error term describes the numerical deviation between the model output power and the target power trend, and is usually measured using mean squared error (MSE). The control response deviation term describes the morphological difference between the voltage response curve generated by the model and the target response curve at various time points, and can be measured using weighted absolute error or time-dependent deviation index. The weight parameters between the two losses are determined according to the importance of power output accuracy and voltage response stability in the application scenario, and are set as adjustable coefficients α and β. In the joint loss function, it is expressed as: Loss_total=α×Loss_power+β×Loss_response, where α and β can be optimized and adjusted through cross-validation or hyperparameter search strategies to balance the model's performance on the two performance indicators and ensure that it can converge synchronously to the target solution that takes into account both prediction accuracy and response accuracy during training.

[0048] Specifically, in each training iteration, the input sequence is fed into the model to be trained to obtain the output power sequence and the corresponding voltage response sequence. Then, the mean square error of the output power sequence and the known target power change trend is calculated as the power prediction error term. At the same time, the output voltage response sequence is compared with the target response curve to calculate the amplitude difference and response delay at key time nodes, thus obtaining the control response deviation term. Then, the two loss values ​​are weighted and summed according to the pre-set weight parameters to form the total loss function. Backpropagation is performed on the total loss function. Based on the calculated gradient information, the model parameters are updated using optimizers such as Adam or SGD. The above operations are repeated until the error convergence criterion is reached on both the training set and the validation set, thereby completing the entire joint optimization process.

[0049] In one embodiment, such as Figure 3 As shown, in step S10, the dynamic target voltage range generated based on the preset load behavior prediction model includes: S106: Construct a multi-dimensional input sequence of voltage and current information according to a preset time sequence, and input it into a preset load behavior prediction model to obtain the power change trend of the target power supply device within the target prediction period.

[0050] In this embodiment, the preset time sequence refers to the time scale arrangement order set according to the operating cycle and data sampling frequency of the target power device.

[0051] Specifically, firstly, raw voltage and current information data are collected from the target power supply device over a continuous operating period. This information is then time-aligned according to the actual collection timestamp and arranged into a two-dimensional data array in chronological order. Each time step corresponds to a set of voltage and current values, constructed as [[U1,I1],[U2,I2],…,[U n ,I n The input sequence structure is then used, and according to the set prediction period, such as the next 10 time steps, the input sequence is fed into the trained load behavior prediction model. The model then generates the power change trend within the corresponding prediction period based on the existing input. S107: Based on the power change trend and combined with the historical steady-state voltage output range, perform prediction offset correction to obtain the expected dynamic output voltage range.

[0052] In this embodiment, the historical steady-state voltage output range refers to the range of voltage output values ​​recorded by the target power supply device during stable operation under multiple typical load scenarios. This range is determined by the upper and lower limits of voltage fluctuations of the device under various stable load conditions.

[0053] Specifically, firstly, steady-state operating segments with power change rates within a set threshold are selected from historical load operation data. Within each steady-state segment, the highest and lowest voltage output values ​​are extracted, and cluster analysis or statistical analysis is performed on the upper and lower bounds of multiple segments. For example, the average maximum and minimum values ​​are calculated, or the 95% confidence interval boundary is taken as the historical steady-state voltage output range for this type of load scenario. Then, the predicted power change trend is mapped to the possible corresponding voltage output level. By consulting the historical steady-state range corresponding to this power, the abnormal offset in the predicted voltage path is compared with the steady-state boundary. If there is an offset that exceeds the steady-state range, it is trimmed or compressed by linear regression according to the upper and lower boundaries of the steady state, thereby obtaining a dynamic output voltage range that conforms to the load response law.

[0054] S108: Perform upper and lower boundary calibration on the dynamic output voltage range to generate a dynamic target voltage range that meets the stable response requirements of the current load behavior.

[0055] Specifically, the electrical response characteristic parameters corresponding to the current load behavior are first identified. These parameters include, but are not limited to, the current response rate under unit voltage change, the load power stabilization time under voltage disturbance, and dynamic response indicators such as response delay within a continuous control cycle. Based on these characteristic parameters, a matching voltage change amplitude range is selected as the initial reference range. Then, the upper and lower boundaries of the dynamic output voltage range are calibrated. The calibration operation includes comparing the differences in amplitude and rate of change between the initial reference range and the dynamic output voltage range. If the dynamic output voltage range exceeds the upper boundary of the reference range, a nonlinear compression processing is performed on the upper boundary using an amplitude compression algorithm. For example, an exponential decay function is used to compress the excessively high voltage amplitude to near the upper limit of the stable response that the load can withstand. If the lower boundary is lower than the load response threshold, a smooth increase strategy is used to raise the voltage. For example, a certain linear adjustment factor is applied near the boundary value to avoid triggering the load anomaly protection mechanism. Finally, the calibrated upper and lower boundaries are combined to form the dynamic target voltage range under the current load behavior.

[0056] Furthermore, amplitude compression algorithm refers to a numerical adjustment method based on nonlinear functions, which is used to control the voltage change amplitude within a stable and controllable dynamic range without causing abnormal load response.

[0057] If the upper boundary of the current voltage is V max The maximum allowable value for a stable load response is V. safe Then, the following processing can be achieved using the exponential compression function: V adjusted =V safe +(V max -V safe )·e -λ·Δt , where Vadjusted λ represents the upper boundary of the compressed voltage, λ is the compression ratio parameter, and Δt is the control time span from the current boundary value setpoint.

[0058] A smooth ramp-up strategy is a control method used to adjust the lower boundary value of a dynamic output voltage range. Based on linear or nonlinear increasing functions, such as slope control functions, logarithmic functions, or piecewise multi-order ramp curves, the lower boundary value V of the original voltage is adjusted. min Perform dynamic elevation processing. The expression is as follows: V lower (t)=V min +α·f(t), where V lower f(t) represents the dynamic lower boundary at a certain moment, α is the rise coefficient, f(t) is a smooth function that satisfies the monotonically increasing characteristic, and t is the current control time step.

[0059] In one embodiment, such as Figure 4 As shown, in step S20, which involves constructing a joint control weight vector based on power deviation information, the following steps are included: S201: Perform sliding statistical processing on the power deviation information within a preset short time window and a preset long time window respectively, and extract short-term fluctuation features and long-term trend features.

[0060] Specifically, within a preset short-term window and a preset long-term window, time periods of the power deviation information sequence are selected for sliding statistical processing. The short-term window uses a smaller sampling length to capture the abrupt changes and fluctuation frequency of the power deviation information within a local time. The standard deviation of the power deviation value within this window and the mean difference between adjacent sampling points are calculated and extracted as short-term fluctuation features. The long-term window uses a larger sampling interval covering multiple periods to analyze the direction of change and steady-state shift of the power deviation information throughout the entire observation period. The trend curve of the power deviation information and the time series within this window is linearly fitted, and the fitting slope and fitting residual are extracted as long-term trend features. The entire feature extraction process is performed independently in the two time windows without cross-processing, ensuring that the short-term fluctuation features and long-term trend features can reflect the power deviation behavior characteristics at different scales.

[0061] S202: Determine the voltage regulation priority factor based on short-term fluctuation characteristics, and determine the current regulation sensitivity factor based on long-term trend characteristics.

[0062] Specifically, when determining the voltage regulation priority factor based on short-term fluctuation characteristics, the standard deviation index in the short-term fluctuation characteristics is first subjected to interval mapping processing. The numerical range is divided into multiple voltage regulation level segments according to a preset ratio. When the standard deviation is in the high fluctuation segment, a higher regulation priority factor value is assigned, and vice versa. Furthermore, the mapping from standard deviation to priority factor can be achieved by setting an exponential or piecewise linear mapping function to enhance the nonlinear response capability. When determining the current adjustment sensitivity factor based on long-term trend characteristics, the fitted slope is first subjected to sign discrimination and absolute value normalization processing. A positive slope represents a continuous upward trend in power, and a negative slope represents a continuous downward trend in power. After evaluating the fitting reliability in combination with the residual size, the corresponding sensitivity factor value is determined by looking up a table or calculating a function. The specific value of the sensitivity factor is used to characterize the amplification of the trend response change when adjusting the current, thereby dynamically adjusting the control sensitivity to the current change amplitude in the control strategy.

[0063] S203: Combine the voltage regulation priority factor and the current regulation sensitivity factor according to a weighted strategy to generate a joint regulation weight vector.

[0064] Specifically, when generating a joint control weight vector by combining the voltage regulation priority factor and the current regulation sensitivity factor using a weighted strategy, a set of weight coefficients for weighted combination is first set. The weight value of each factor is determined by consulting historical control effect data or based on prior control experience. On this basis, the control weight of each time step is calculated using a linear weighting method. The joint control weight is equal to the voltage regulation priority factor multiplied by its corresponding weight plus the current regulation sensitivity factor multiplied by its corresponding weight. Then, the generated control weight is normalized or truncated according to the target control accuracy or equipment stability requirements to ensure that the final output joint control weight vector meets the amplitude and resolution requirements of the input-to-response model at all control time steps. For example, if the voltage regulation demand is higher than the current regulation demand in a certain control period, the weighting coefficient of the voltage factor can be increased to 0.7 and the weighting coefficient of the current factor can be decreased to 0.3. The final joint control weight vector reflects the control priority of voltage change in that stage.

[0065] In one embodiment, such as Figure 5 As shown, in step S30, the target power evolution trajectory curve is constructed using a physical response inversion mechanism based on the joint control weight vector and the dynamic target voltage range, including: S301: Based on the joint control weight vector, calculate the voltage change and current response amplitude corresponding to each stage in the voltage regulation path, and generate the voltage change path based on the dynamic target voltage range.

[0066] Specifically, when calculating the voltage change and current response amplitude corresponding to each stage in the voltage regulation path based on the joint control weight vector, the target power evolution trajectory curve is first segmented according to a preset time step. For each time step, the corresponding control weight value is read and combined with the upper and lower boundary intervals of the current period in the dynamic target voltage range. The direction and absolute value of the voltage adjustment within that time step are calculated using interpolation or piecewise linear functions. For example, when the joint control weight of a certain time step is high and the corresponding power target shows an upward trend, the voltage change is calculated by adjusting the voltage stepwise along the upper boundary direction starting from the current voltage value. Then, combined with a preset load voltage-current response model, such as an exponential response model or a fitted impedance characteristic function, the voltage change is input into the model to derive the corresponding current response amplitude, which is recorded as the voltage-current pair for that time step. Finally, all voltage changes are connected in the order of the time steps to form a voltage change path that is consistent with the target control direction and has a controlled amplitude.

[0067] Furthermore, the load voltage-current response model is preferably a linear approximate impedance model. This model is based on the characteristic that the actual load exhibits an approximately linear response within the local steady-state range. By setting the complex impedance Z = R + jX, the impedance is approximately linear within each control time step. The current response amplitude is derived in the form of a formula, where the voltage change V comes from the matching interpolation result of the joint control weight vector and the dynamic target voltage range. The R and X values ​​in the model can be set based on historical measurement data or static load nominal values. In actual calculation, the active current can be calculated by separating the real part and the reactive response can be estimated by the imaginary part, so as to quickly obtain the current response amplitude corresponding to the target voltage change path. It is suitable for most medium and low voltage steady-state load simulation processes.

[0068] S302: Based on the preset load electrical response relationship, perform response inversion operation on the voltage change path to obtain the current response path.

[0069] In this embodiment, the preset load electrical response relationship refers to a deterministic physical mapping relationship between the load voltage input and current output. This mapping can be described by a basic circuit model combining resistors, capacitors, and inductors. The resistance component represents the linear proportional relationship between voltage and current, the capacitance component represents the relationship between the rate of change of current and voltage, and the inductance component represents the relationship between the integral of current and voltage. The response inversion operation refers to calculating the current response path by inversely deriving the contribution of resistance, capacitance, and inductance to the current response in the load model based on the known voltage change path. For example, in the equivalent model containing a combined RC inductor load, the relationship between current I(t) and voltage V(t) can be expressed as follows: Where k1, k2, and k3 are the weighting factors for resistive, capacitive, and inductive responses, respectively, and R, C, and L are the equivalent resistance, capacitance, and inductance values ​​of the load, respectively. By substituting the known voltage change path V(t) into the above formula and calculating the contribution values ​​of the proportional, derivative, and integral terms, the current response path on the time series is obtained by superimposing them.

[0070] Specifically, when performing response inversion on the voltage change path based on the preset load electrical response relationship, the numerical inversion operation is performed by analyzing the nonlinear coupling relationship formula between voltage and current according to the voltage change value at each time step and the corresponding load response parameters. In each time step, the transient response is calculated using Ohm's law and the relationship between inductive reactance and inductive reactance. The current value corresponding to the current voltage change is gradually approximated through iteration. The continuity of the results of the previous time step is maintained in the continuous time series to ensure the physical consistency and response stability of the inversion results. For example, when the voltage change is linearly increasing, the current response value under the voltage condition is calculated based on the known load resistance value and inductance change rate, and the obtained current values ​​are arranged in time order to form a complete current response path.

[0071] S303: Multiply the voltage change and the current response amplitude according to the control time step to obtain the initial power evolution trajectory.

[0072] Specifically, the aforementioned voltage change path and current response path are represented as {ΔV1, ΔV2, ..., ΔV} according to the control time step. n} and {ΔI1, ΔI2, ..., ΔI n}, at each control time step t i Within this process, the power change at the current moment, P, is calculated using a product operation. i =ΔV i ×ΔI i , where △V i This represents the voltage change at that moment, ΔI. i This represents the corresponding current response amplitude. The product result indicates the contribution of the current control behavior to the instantaneous load power. By sequentially calculating the power changes from i=1 to n at all times throughout the entire control cycle, the initial power evolution trajectory sequence {P1, P2, ..., P} is obtained. n}

[0073] S304: Compare the initial power evolution trajectory with the target power change trend point by point to extract the power deviation value at each control time step.

[0074] Specifically, within the control period, the initial power evolution trajectory sequence {P1, P2, ..., P...} is... n}, and the target power change trend sequence Perform point-to-point correspondence at each control time step t i The difference between the initial power value and the target power value is calculated and denoted as . This difference reflects the power output deviation caused by the current combination of control parameters. This method is used to perform calculations across all control time steps to obtain the complete power deviation values ​​{ε1, ε2, ..., ε...}. n}

[0075] S305: Based on the power deviation value, adjust the corresponding voltage change and current response amplitude in reverse, and update the voltage change path and current response path simultaneously to obtain the updated voltage change path and updated current response path.

[0076] Specifically, based on the power deviation value ε at each control time step i The control parameters are corrected through a reverse adjustment function, which uses a linear mapping method and is calculated by multiplying each parameter by a set voltage correction coefficient θ. u Using the current correction factor θi, the correction value δU for the voltage change is calculated. i =-θ u ·ε i The correction value δI for the current response amplitude i =-θ i ·ε i Then adjust the voltage change and current response amplitude in the current control time step to ΔU′ respectively. i =ΔU i +δU i and ΔI′ i =ΔI i +δI i This process is continuously executed to progressively correct the control parameters for all control time steps, ultimately reconstructing the updated voltage change path {ΔU′1, ΔU′2, ..., ΔU′...}. n} and current response path {ΔI′1, ΔI′2, ..., ΔI′ n}

[0077] S306: Recalculate the power evolution trajectory based on the updated voltage change path and the updated current response path, and repeatedly compare and correct the power evolution trajectory with the target power change trend until the power deviation in each control time step meets the preset convergence condition, and finally determine the target power evolution trajectory curve.

[0078] Specifically, according to the control time step t i The order is based on the updated voltage change path {ΔU′1, ΔU′2, ..., ΔU′...} n} and the updated current response path {ΔI′1, ΔI′2, ..., ΔI′ n}, perform the product operation P′ at each time step. i =ΔU′ i ·ΔI′ i Recalculate to form a new power evolution trajectory {P′1, P′2, ..., P′} n}, and then compared the power evolution trajectory with the target power change trend. Perform point-by-point interpolation to obtain the power deviation value at each time step. If the maximum deviation value meets the preset convergence threshold ε max If the value is less than or equal to δ, then the current power evolution trajectory is confirmed as the final target power evolution trajectory curve.

[0079] In one embodiment, such as Figure 6 As shown, in step S50, according to the control parameter sequence, instruction encoding processing is performed according to preset hardware and software interface constraints to generate analog control instructions that match the target power evolution trajectory curve, including: S501: Based on the constraints of the software and hardware interface, determine the field position and signal representation format of each control parameter pair in the target control instruction structure.

[0080] Specifically, based on the requirements of the hardware and software interface constraints on the control command format, the definition of the data field structure in the control interface protocol document is first read. By analyzing the logical order and function characteristics of each pair of control parameters in the control parameter sequence, they are mapped to the field areas in the control command structure. For example, parameters used for voltage regulation are assigned to bytes 2 to 5 in the command structure, and parameters used for current regulation are assigned to bytes 6 to 9. Further, referring to the signal encoding methods specified in the interface protocol, such as high-order bit first, two's complement encoding, or fixed-point decimal shift rules, the physical quantity value range and precision bits of each control parameter are matched with the required digital representation. Finally, the signal representation format that should be used for the start and end positions of each control parameter in the field of the control command structure is determined.

[0081] S502: Based on the signal representation format, each pair of control parameters is converted into a recognizable numerical control signal and embedded into the control instruction structure according to the field position to obtain an initial instruction set containing control information at multiple times.

[0082] Specifically, based on the determined signal representation format, the voltage and current change values ​​in each control parameter pair are formatted and converted. For example, when using fixed-point decimal encoding, the real-number control parameters are multiplied by a preset scaling factor to convert them into integer data. Overflow checks and sign bit padding are performed according to the byte length requirements. Then, the converted numerical control signals are embedded into the corresponding byte segments in the instruction structure according to the field position. For example, the formatted voltage change values ​​are written to bytes 2 to 5, and the current change values ​​are written to bytes 6 to 9. The above process is repeated to process the control parameter pairs at each time step, and finally an initial instruction set is obtained, which is arranged in chronological order and each instruction contains a complete control field.

[0083] S503: Serializes the initial instruction set in chronological order and inserts a synchronization flag field according to preset hardware and software interface constraints to generate analog control instructions.

[0084] Specifically, according to the time tags corresponding to each control parameter in the control parameter sequence, the initial instruction set is sorted in ascending order to construct an instruction flow structure that strictly follows the control time step. In the sorted instruction flow, the insertion rhythm and position of the synchronization mark field are determined according to the preset hardware and software interface constraints. For example, a synchronization mark field is inserted every N control instructions or when a critical time node is detected. The synchronization mark field can use a specific width of hexadecimal encoding to represent the synchronization signal edge, rhythm identification code, and control status bit. For example, if an 8-bit structure is used, the first two bits are "01" to indicate rising edge triggering, and the last six bits are a cycle counter or identification mask. The selection of the insertion position needs to be combined with the response cycle requirements of the synchronization timing in the interface specification. Finally, the above-mentioned synchronization mark field is inserted into the original instruction sequence to complete the generation of analog control instructions with complete format and coherent timing.

[0085] In one embodiment, such as Figure 7 As shown, in step S503, the initial instruction set is serialized according to time sequence, and a synchronization flag field is inserted according to preset hardware and software interface constraints to generate analog control instructions, including: S5031: Sort the time tags of each control parameter pair in the initial instruction set in ascending order to obtain a time-sequenced instruction stream.

[0086] Specifically, for each pair of control parameters in the initial instruction set, the corresponding time stamp information is read. The time stamp can be represented as a timestamp value in milliseconds or microseconds. A stable time base, such as the system startup time or reference trigger point, is used as the starting time reference. All control parameter pairs and time stamps are combined into a set of tuples, and a stable sorting algorithm, such as merge sort or heap sort, is used to sort the time stamps in the set in ascending order. During the sorting process, the original order of control parameter pairs under the same time stamp is maintained to ensure that the output sequence is strictly monotonically increasing in time and has a consistent logical order. After sorting, each control parameter pair is arranged in order according to the sorting result to form a time-seriesd instruction stream.

[0087] S5032: Based on the preset hardware and software interface constraints, determine the insertion strategy of the synchronization marker field in the instruction stream and generate an insertion position list.

[0088] Specifically, by analyzing parameters such as data processing cycle, signal recognition accuracy, and synchronization response delay in the hardware and software interface constraints, the position range where synchronization marker fields need to be inserted in the control command stream is calculated. The insertion strategy of the synchronization marker field can be divided into two categories according to the control rhythm: the equal-periodic insertion strategy and the event-driven insertion strategy. The equal-periodic insertion strategy refers to segmenting the command stream at fixed time intervals, and inserting the synchronization marker field at the beginning of each segment. The time interval is calculated based on the minimum command parsing cycle specified by the interface. The event-driven insertion strategy refers to inserting the synchronization marker field based on key event nodes (such as voltage transition points, current change points, or control switching points). Event nodes can be identified by analyzing the step amplitude of voltage change values ​​and current change values ​​in the time series, which exceeds a set threshold. The command stream index number corresponding to the time tag that meets the above conditions is recorded as the insertion position, and finally a list of synchronization marker field insertion positions is formed.

[0089] S5033: In the insertion position list, embed a synchronization mark field containing rhythm identification identifier and control synchronization flag, and constrain the field structure and encoding method of the synchronization mark field to form a simulated control command.

[0090] Specifically, based on the time index marked in the insertion position list, the synchronization marker field is embedded into the control instruction stream at the corresponding position. Each synchronization marker field includes a rhythm identification identifier and a control synchronization flag. The rhythm identification identifier uses fixed-length binary encoding to represent the control rhythm level. For example, 00 can represent a low-speed rhythm, 01 can represent a medium-speed rhythm, and 10 can represent a high-speed rhythm. The control synchronization flag is used to mark whether it is the start point of the control instruction cycle. Its field value can be set to 1 to indicate the start of synchronization and 0 to indicate the relay state. The field structure adopts a TLV (Type-Length-Value) structure, that is, the field type and field length are defined first, and then the corresponding values ​​are filled in to ensure that the interface parsing module can accurately identify and execute. In order to avoid exceeding the instruction frame length limit, the synchronization marker field must meet the interface's limit on the maximum number of bits per frame when encoding. A compression encoding method can be used to merge the rhythm identifier and the synchronization flag into a single-byte instruction. Finally, after embedding all the synchronization marker fields, a complete and rhythmically aligned analog control instruction stream is formed.

[0091] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0092] In one embodiment, a load power trajectory simulation control device is provided, which corresponds one-to-one with the load power trajectory simulation control method described in the above embodiments. For example... Figure 8 As shown, this load power trajectory simulation control device includes an electrical parameter acquisition and power deviation calculation module, a dynamic target voltage generation module, a joint control weight construction module, a physical response-driven trajectory reconstruction module, and a simulation control command generation module. Detailed descriptions of each functional module are as follows: The electrical parameter acquisition and power deviation calculation module is used to acquire voltage and current information of the target power supply equipment, determine power deviation information, and generate a dynamic target voltage range based on a preset load behavior prediction model. The dynamic target voltage generation module is used to construct a joint control weight vector based on power deviation information; The joint control weight construction module is used to construct the target power evolution trajectory curve based on the joint control weight vector and the dynamic target voltage range using the physical response inversion mechanism. The physical response-driven trajectory reconstruction module is used to perform time discretization processing on the target control trajectory curve and extract the voltage change value and current change value corresponding to each time step. The analog control command generation module is used to combine voltage and current change values ​​in time sequence to form a control parameter sequence. Based on the control parameter sequence, the module performs command encoding processing according to preset hardware and software interface constraints to generate analog control commands that match the target power evolution trajectory curve.

[0093] Optional, the electrical parameter acquisition and power deviation calculation module includes: The load scenario historical electrical parameter acquisition submodule is used to acquire historical voltage and current information under multiple typical load scenarios; The timing input construction and target information extraction submodule is used to construct an input sequence from historical voltage and current information in chronological order, and to obtain the target power change trend and control response deviation information corresponding to the input sequence. The joint training sample construction submodule is used to construct a joint training sample set based on the input sequence, the target power change trend, and the control response deviation information. The prediction error and response deviation calculation submodule is used to train the model parameters using the joint training sample set. During the training process, the power prediction error information is calculated based on the difference between the predicted power value output by the model and the target power change trend, and the control response deviation information is calculated based on the error value between the voltage response curve generated by the model and the target response curve. The multi-objective loss optimization and model building submodule is used to jointly optimize power prediction error information and control response deviation information based on a preset weighted multi-objective loss function, so as to obtain a preset load behavior prediction model.

[0094] Optionally, the electrical parameter acquisition and power deviation calculation module also includes: The timing input construction submodule is used to construct a multi-dimensional input sequence of voltage and current information according to a preset time sequence, and input it into a preset load behavior prediction model to obtain the power change trend of the target power supply device within the target prediction period; the voltage range prediction correction submodule is used to perform prediction offset correction based on the power change trend and the historical steady-state voltage output range to obtain the expected dynamic output voltage range. The dynamic voltage range generation submodule is used to calibrate the upper and lower boundaries of the dynamic output voltage range and generate a dynamic target voltage range that meets the requirements for stable response to the current load behavior.

[0095] Optionally, the dynamic target voltage generation module includes: The power deviation feature extraction submodule is used to perform sliding statistical processing on the power deviation information within a preset short time window and a preset long time window respectively, and extract short-term fluctuation features and long-term trend features. The regulation factor generation submodule is used to determine the voltage regulation priority factor based on short-term fluctuation characteristics and the current adjustment sensitivity factor based on long-term trend characteristics. The joint control weight construction submodule is used to combine the voltage regulation priority factor and the current adjustment sensitivity factor according to the weighting strategy to generate a joint control weight vector.

[0096] Optionally, the joint regulation weight construction module includes: The regulation path construction submodule is used to calculate the voltage change and current response amplitude corresponding to each stage in the voltage regulation path according to the joint regulation weight vector, and generate the voltage change path based on the dynamic target voltage range. The current response inversion submodule is used to perform response inversion operations on the voltage change path based on the preset load electrical response relationship to obtain the current response path; The initial trajectory generation submodule is used to multiply the voltage change and the current response amplitude according to the control time step to obtain the initial power evolution trajectory; The error extraction and feedback correction submodule is used to compare the initial power evolution trajectory with the target power change trend point by point and extract the power deviation value at each control time step. The current response path update submodule is used to reversely adjust the corresponding voltage change and current response amplitude according to the power deviation value, and synchronously update the voltage change path and current response path to obtain the updated voltage change path and updated current response path. The trajectory iteration optimization submodule is used to recalculate the power evolution trajectory based on the updated voltage change path and the updated current response path, and repeatedly compare and correct the power evolution trajectory with the target power change trend until the power deviation in each control time step meets the preset convergence condition, and finally determine the target power evolution trajectory curve.

[0097] Optionally, the analog control command generation module includes: The instruction field mapping submodule is used to determine the field position and signal representation format of each control parameter pair in the control parameter sequence in the target control instruction structure based on the constraints of the software and hardware interface. The instruction embedding and encoding submodule is used to convert each pair of control parameters into recognizable numerical control signals based on the signal representation format, and embed them into the control instruction structure according to the field position to obtain an initial instruction set containing control information at multiple times. The serialization and synchronization generation submodule is used to serialize the initial instruction set in chronological order and insert synchronization marker fields according to preset hardware and software interface constraints to generate analog control instructions.

[0098] Optional, the serialization synchronization generation submodule includes: The instruction timing reordering submodule is used to sort the time tags of each control parameter pair in the initial instruction set in ascending order to obtain a time-sequenced instruction stream. The synchronization field insertion strategy submodule is used to determine the insertion strategy of the synchronization marker field in the instruction stream based on preset software and hardware interface constraints, and generate a list of insertion positions. The synchronization field embedding and constraint configuration submodule is used to embed a synchronization mark field containing a rhythm identification identifier and a control synchronization flag in the insertion position list, and to constrain and configure the field structure and encoding method of the synchronization mark field to form a simulated control command.

[0099] For specific limitations regarding a load power trajectory simulation control device, please refer to the limitations of a load power trajectory simulation control method described above, which will not be repeated here. Each module in the aforementioned load power trajectory simulation control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0100] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a load power trajectory simulation control method.

[0101] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Acquire voltage and current information of the target power supply device, determine power deviation information, and generate a dynamic target voltage range based on a preset load behavior prediction model; Based on power deviation information, a joint control weight vector is constructed; Based on the joint control weight vector and the dynamic target voltage range, the target power evolution trajectory curve is constructed using the physical response inversion mechanism; The target control trajectory curve is discretized over time to extract the voltage and current changes at each time step. The voltage and current changes are then combined in chronological order to form a control parameter sequence. Based on the control parameter sequence, the command is encoded according to preset hardware and software interface constraints to generate a simulated control command that matches the target power evolution trajectory curve.

[0102] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Acquire voltage and current information of the target power supply device, determine power deviation information, and generate a dynamic target voltage range based on a preset load behavior prediction model; Based on power deviation information, a joint control weight vector is constructed; Based on the joint control weight vector and the dynamic target voltage range, the target power evolution trajectory curve is constructed using the physical response inversion mechanism; The target control trajectory curve is discretized over time to extract the voltage and current changes at each time step. The voltage and current changes are then combined in chronological order to form a control parameter sequence. Based on the control parameter sequence, the command is encoded according to preset hardware and software interface constraints to generate a simulated control command that matches the target power evolution trajectory curve.

[0103] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0104] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0105] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A load power trajectory simulation control method, characterized in that, The load power trajectory simulation control method includes: Acquire voltage and current information of the target power supply device, determine power deviation information, and generate a dynamic target voltage range based on a preset load behavior prediction model; Based on the power deviation information, a joint control weight vector is constructed; Based on the joint control weight vector and the dynamic target voltage range, the target power evolution trajectory curve is constructed using the physical response inversion mechanism; Perform time discretization processing on the target control trajectory curve to extract the voltage change value and current change value corresponding to each time step; The voltage change value and the current change value are combined in time sequence to form a control parameter sequence. Based on the control parameter sequence, instruction encoding is performed according to preset hardware and software interface constraints to generate an analog control instruction that matches the target power evolution trajectory curve.

2. The load power trajectory simulation control method according to claim 1, characterized in that, The preset load behavior prediction model previously included: Acquire historical voltage and current information under multiple typical load scenarios; The historical voltage information and the historical current information are arranged in chronological order to construct an input sequence, and the target power change trend and control response deviation information corresponding to the input sequence are obtained. Based on the input sequence, the target power change trend, and the control response deviation information, a joint training sample set is constructed; The parameters of the model are trained using the joint training sample set. During the training process, the power prediction error information is calculated based on the difference between the predicted power value output by the model and the target power change trend. The control response deviation information is calculated based on the error between the voltage response curve generated by the model and the target response curve. Based on a preset weighted multi-objective loss function, the power prediction error information and the control response deviation information are jointly optimized to obtain the preset load behavior prediction model.

3. The load power trajectory simulation control method according to claim 1, characterized in that, The dynamic target voltage range generated based on the preset load behavior prediction model includes: The voltage information and the current information are constructed into a multi-dimensional input sequence according to a preset time sequence, and then input into the preset load behavior prediction model to obtain the power change trend of the target power supply device within the target prediction period. Based on the power change trend, a prediction offset correction is performed in conjunction with the historical steady-state voltage output range to obtain the expected dynamic output voltage range. The upper and lower boundaries of the dynamic output voltage range are calibrated to generate the dynamic target voltage range that meets the stable response requirements of the current load behavior.

4. The load power trajectory simulation control method according to claim 1, characterized in that, The construction of a joint control weight vector based on the power deviation information includes: Sliding statistical processing is performed on the power deviation information within a preset short time window and a preset long time window respectively to extract short-term fluctuation features and long-term trend features. The voltage regulation priority factor is determined based on the short-term fluctuation characteristics, and the current regulation sensitivity factor is determined based on the long-term trend characteristics. The voltage regulation priority factor and the current regulation sensitivity factor are combined according to a weighted strategy to generate the joint regulation weight vector.

5. The load power trajectory simulation control method according to claim 1, characterized in that, The step of constructing the target power evolution trajectory curve using a physical response inversion mechanism based on the joint control weight vector and the dynamic target voltage range includes: Based on the joint control weight vector, the voltage change and current response amplitude corresponding to each stage in the voltage regulation path are calculated, and a voltage change path is generated based on the dynamic target voltage range. Based on the preset load electrical response relationship, a response inversion operation is performed on the voltage change path to obtain the current response path; The voltage change and the current response amplitude are multiplied by the control time step to obtain the initial power evolution trajectory. The initial power evolution trajectory is compared with the target power change trend point by point to extract the power deviation value at each control time step; Based on the power deviation value, the corresponding voltage change and current response amplitude are adjusted in reverse, and the voltage change path and current response path are updated synchronously to obtain the updated voltage change path and updated current response path. The power evolution trajectory is recalculated based on the updated voltage change path and the updated current response path. The power evolution trajectory is then repeatedly compared and corrected with the target power change trend until the power deviation in each control time step meets the preset convergence condition. Finally, the target power evolution trajectory curve is determined.

6. The load power trajectory simulation control method according to claim 1, characterized in that, The step of generating analog control commands that match the target power evolution trajectory curve by performing instruction encoding processing according to the control parameter sequence and preset software and hardware interface constraints includes: Based on the hardware and software interface constraints, determine the field position and signal representation format of each control parameter pair in the control parameter sequence within the target control instruction structure; Based on the signal representation format, each pair of control parameters is converted into a recognizable numerical control signal and embedded into the control instruction structure according to the field position to obtain an initial instruction set containing multi-time control information. The initial instruction set is serialized in chronological order, and a synchronization flag field is inserted according to the preset hardware and software interface constraints to generate the simulation control instructions.

7. A load power trajectory simulation control method according to claim 6, characterized in that, The step of serializing the initial instruction set in chronological order and inserting a synchronization flag field according to the preset hardware and software interface constraints to generate the simulation control instructions includes: Arrange the time tags of each control parameter pair in the initial instruction set in ascending order to obtain a time-sequenced instruction stream; Based on the preset hardware and software interface constraints, the insertion strategy of the synchronization marker field in the instruction stream is determined, and an insertion position list is generated; In the insertion position list, a synchronization marker field containing a rhythm identification identifier and a control synchronization flag is embedded. The field structure and encoding method of the synchronization marker field are constrained and configured to form the analog control command.

8. A load power trajectory simulation control device, characterized in that, The load power trajectory simulation control device includes: The electrical parameter acquisition and power deviation calculation module is used to acquire voltage and current information of the target power supply device, determine power deviation information, and generate a dynamic target voltage range based on a preset load behavior prediction model. A dynamic target voltage generation module is used to construct a joint control weight vector based on the power deviation information; The joint control weight construction module is used to construct the target power evolution trajectory curve based on the joint control weight vector and the dynamic target voltage range using a physical response inversion mechanism. The physical response-driven trajectory reconstruction module is used to perform time discretization processing on the target control trajectory curve and extract the voltage change value and current change value corresponding to each time step. The analog control command generation module is used to combine the voltage change value and the current change value in time sequence to form a control parameter sequence, and perform command encoding processing according to the control parameter sequence and preset software and hardware interface constraints to generate an analog control command that matches the target power evolution trajectory curve.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the load power trajectory simulation control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the load power trajectory simulation control method as described in any one of claims 1 to 7.

Images