Testing method of three-phase primary co-current and co-voltage simulation on-load relay protection vector inspection device

Abstract

The invention relates to the technical field of load testing, in particular to a testing method of a three-phase primary co-current and co-voltage simulation on-load relay protection vector inspection device, which comprises the following steps: acquiring a fundamental wave parameter, and performing chaotic modulation on a dynamic frequency modulation signal according to the fundamental wave parameter to obtain a frequency modulation voltage signal; tracking a fundamental wave of a target load through the frequency modulation voltage signal according to a phase-locked loop to obtain a fundamental wave phase; and performing load impedance optimization on a target load through the frequency modulation voltage signal according to deep reinforcement learning to obtain the optimal load impedance and the optimal power factor of the target load. According to the method, the power factor of the load is optimized through reinforcement learning, the consumption of reactive power is reduced, so that the efficiency of a power system is improved, and the optimized power factor can reduce the loss of the system, improve the utilization rate of power equipment and reduce the burden on a power grid.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of load test, in particular to a three-phase primary same-flow same-pressure simulation load relay protection vector checking device test method. BACKGROUND

[0002] The traditional method usually relies on static measurement and fixed value setting, which leads to the inability to accurately track and adjust the fundamental parameters in real time when the load changes dynamically. This static method may not adapt to the complex fluctuations of the load, resulting in a large measurement error. Especially in a high dynamic environment, it may not respond to load changes in time. In the traditional method, the impedance optimization of the load often relies on manual setting or a simple algorithm, lacking self-adaptive ability. For rapid changes in load, the traditional method may not be able to adjust the load impedance in time, thereby affecting the overall efficiency of the power system, increasing energy loss, and reducing the stability of the power system. In the traditional method, the power factor is usually adjusted by a relatively simple control means, lacking an intelligent optimization process. The traditional method may not automatically adapt to load changes, resulting in the inability to effectively reduce the consumption of reactive power, and the efficiency of the power system and the utilization rate of the equipment cannot be maximized. In addition, the traditional method usually uses frequency domain analysis or single-scale time domain analysis, lacking comprehensive capture of signal details. The traditional method often cannot effectively detect low-frequency or high-frequency noise and harmonic components, resulting in a blind area in analyzing the state of the power system, making it difficult to identify potential risks and problems in time. SUMMARY

[0003] The technical problem to be solved by the present application is to overcome the shortcomings of the above-mentioned prior art and provide a three-phase primary same-flow same-pressure simulation load relay protection vector checking device test method.

[0004] The technical solution adopted to solve the above technical problems is: a three-phase primary same-flow same-pressure simulation load relay protection vector checking device test method, comprising: obtaining a fundamental parameter, and performing chaotic modulation on a dynamic frequency modulation signal according to the fundamental parameter to obtain a frequency modulation voltage signal; tracking the fundamental wave of a target load through the frequency modulation voltage signal according to a phase-locked loop to obtain a fundamental wave phase; optimizing the load impedance of the target load through the frequency modulation voltage signal according to deep reinforcement learning to obtain the optimal load impedance and the optimal power factor of the target load; performing real-time signal acquisition on the target load with the optimal load impedance and the optimal power factor to obtain real-time voltage signals and real-time current signals; performing vector analysis on the real-time voltage signal and the real-time current signal according to a multi-scale wavelet transform to obtain a fundamental phase difference and a harmonic energy distribution of the target load; performing protection verification on the target load according to the fundamental phase difference and the harmonic energy distribution to obtain a test label of the target load.

[0005] Preferably, performing chaotic modulation on a dynamic frequency modulation signal according to the fundamental parameter to obtain a frequency modulation voltage signal, comprising: generating a frequency modulation index according to a Logistic chaotic mapping; combining the frequency modulation index with a multi-harmonic injection to generate the frequency modulation voltage signal, wherein a calculation formula of the frequency modulation voltage signal is as follows: ; wherein, denotes the frequency modulation voltage signal, denotes a fundamental voltage, denotes a fundamental frequency, denotes the frequency modulation index, denotes a harmonic random factor, simulating harmonic fluctuation, denotes a random initial phase of the harmonic, denotes an angular frequency.

[0006] Preferably, tracking a fundamental wave of a target load according to a phase-locked loop through the frequency modulation voltage signal to obtain a fundamental phase, comprising: obtaining an output phase of the phase-locked loop, determining a phase error according to the frequency modulation voltage signal and the output phase, and deriving the phase error to obtain a derivative of the phase error, wherein a calculation formula of the phase error is as follows: ; wherein, denotes the phase error, denotes the output phase of the phase-locked loop; when the phase error is greater than a preset first error threshold and the derivative of the phase error is positive, linearly increasing a proportional gain of the phase-locked loop according to a preset increasing weight; when the phase error is less than a preset second error threshold and the derivative of the phase error is negative, linearly decreasing an integral gain of the phase-locked loop according to a preset decreasing weight; tracking the fundamental wave of the target load according to the phase-locked loop to obtain the fundamental phase.

[0007] Preferably, optimizing a load impedance of a target load according to a deep reinforcement learning through the frequency modulation voltage signal to obtain an optimal load impedance and an optimal power factor of the target load, comprising: Define a state space, wherein the state space includes the frequency modulation voltage and the current power factor of the target load; Define an action space, wherein the action space includes resistance adjustment amount and reactance adjustment amount; Define a reward function, wherein the reward function is defined based on the adjustment amount of the current power factor, the target power factor, and the load impedance; Measure the current voltage and power factor, and adjust the load impedance accordingly; By adjusting the resistance and reactance, the load impedance is updated to obtain a new voltage and a new power factor; The reward is calculated based on the current state and the adjusted load impedance, wherein the experience includes state, action, reward and new state, and the experience is stored in the experience replay buffer; A batch of data is sampled from the experience replay buffer, the Critic network is updated by minimizing the error, the Actor network is updated using policy gradient ascent, and the current round is terminated when the maximum number of steps is reached or the power factor stabilizes.

[0008] Preferably, the reward function is as follows: ;in, Represents the reward function, The penalty weight represents the power factor error. This represents the penalty weight for the load impedance adjustment magnitude. This indicates the adjustment amount of the load impedance. Indicates the current power factor. Indicates the target power factor; The update formula for the load impedance is as follows: ;in, Indicates the first Load impedance at the next iteration Indicates the resistance adjustment amount. This indicates the reactance adjustment amount.

[0009] Preferably, vector analysis is performed on the real-time voltage signal and the real-time current signal according to multi-scale wavelet transform to obtain the fundamental phase difference and harmonic energy distribution of the target load, including: The scale parameters are determined based on the sampling frequency and the actual frequency. The translation parameters are determined based on the sampling frequency; The voltage wavelet coefficients corresponding to the real-time voltage signal and the current wavelet coefficients corresponding to the real-time current signal are determined based on the scale parameter and the translation parameter. The foundation dimensions are determined based on the power frequency corresponding to the fundamental wave. extracting a maximum energy ridge of the base scale according to the voltage wavelet coefficients; extracting a first instantaneous phase corresponding to the real-time voltage signal and a second instantaneous phase of the real-time current signal at the maximum energy ridge; removing the base scale, selecting other harmonic scales, and calculating an energy time-frequency distribution of each harmonic scale.

[0010] Preferably, the scale parameter is calculated according to the following formula: ; wherein, represents the scale parameter, represents the center frequency of the wavelet in the multi-scale wavelet transform, represents the sampling frequency, represents the actual frequency; The calculation formula of the translation parameter is as follows: ; wherein, represents the translation parameter; The calculation formula of the base scale is as follows: ; wherein, represents the base scale, represents the fundamental frequency; The calculation formula of the maximum energy ridge is as follows: ; wherein, represents the maximum energy ridge, represents the voltage wavelet coefficient.

[0011] Preferably, the calculation formula of the first instantaneous phase is as follows: ; wherein, represents the first instantaneous phase; The calculation formula of the second instantaneous phase is as follows: ; wherein, represents the first instantaneous phase; The calculation formula of the energy time-frequency distribution is as follows: ; wherein, represents the energy time-frequency distribution, represents the total length of the real-time voltage signal.

[0012] Preferably, according to the fundamental wave phase difference and the harmonic energy distribution, the target load is protected and verified to obtain a test label of the target load, comprising: comparing the fundamental wave phase difference with a preset phase difference threshold, and comparing the harmonic energy distribution with a preset energy distribution threshold; If the fundamental phase difference is greater than a preset phase difference threshold, or the harmonic energy distribution is greater than a preset energy distribution threshold, then the target load is determined to have harmonic interference; otherwise, the target load is determined not to have harmonic interference.

[0013] The beneficial effects of the present invention are as follows: (1) The present invention uses chaotic modulation of dynamic frequency modulation signal to obtain more accurate measurement and tracking of fundamental parameters. Especially under load changes, dynamic frequency modulation signal helps to better capture the characteristics of target load, avoids measurement errors that may exist in traditional static methods, and optimizes the impedance of target load through deep reinforcement learning, which can automatically adjust the impedance of load, making the power system more efficient. The optimized load impedance can improve power transmission efficiency, reduce energy loss, and ensure that the power system is more stable; (2) The present invention optimizes the power factor of load through reinforcement learning, which helps to reduce reactive power consumption, thereby improving the efficiency of power system. The optimized power factor can reduce system losses, improve the utilization rate of power equipment, reduce the burden on the power grid, and through real-time voltage signal The acquisition of current signals provides rich data support for the system, which can reflect the working status of the system in a timely manner. Combined with the time-frequency analysis of the signal by multi-scale wavelet transform, it can detect the detailed changes in the signal more meticulously, such as the fundamental phase difference and harmonic components, and effectively identify potential problems. (3) The present invention can decompose the signal at different scales through wavelet transform, thereby performing multi-level analysis of the signal. This analysis can help to better identify the harmonic components and their energy distribution in the signal, thereby revealing the potential risks and problems of the system, such as harmonic pollution and power quality problems. Furthermore, the fundamental phase difference and harmonic energy distribution obtained based on vector analysis can be used to monitor the health status of the power system in real time. Combined with the protection verification method, potential faults or abnormalities, such as overload and short circuit, can be detected in advance, and protection measures can be taken in time to prevent accidents from occurring. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the overall method steps in one embodiment of the present invention. Detailed Implementation

[0015] Example 1, as Figure 1 As shown, the test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device proposed in this invention includes: S1. Obtain the fundamental wave parameters, and perform chaotic modulation of the dynamic frequency modulation signal based on the fundamental wave parameters to obtain the frequency modulation voltage signal; S2. The phase-locked loop tracks the fundamental wave of the target load through the frequency-modulated voltage signal to obtain the fundamental wave phase. S3, optimizing the load impedance of the target load by a frequency-modulated voltage signal according to deep reinforcement learning, so as to obtain the optimal load impedance and the optimal power factor of the target load; S4, collecting real-time signals of the target load with the optimal load impedance and the optimal power factor in real time, so as to obtain real-time voltage signals and real-time current signals; S5, performing vector analysis on the real-time voltage signals and the real-time current signals according to multi-scale wavelet transform, so as to obtain the fundamental phase difference and the harmonic energy distribution of the target load; S6, verifying the protection of the target load according to the fundamental phase difference and the harmonic energy distribution, so as to obtain a test label of the target load.

[0016] In the application, the fundamental wave refers to the most basic frequency component in the alternating current signal, that is, the main frequency of the signal, and the fundamental wave parameters usually include amplitude, frequency and phase, etc. These parameters are used to describe the main working state in the power system; the chaotic modulation refers to modulating other signals by chaotic signals (usually complex and unpredictable signals) to generate dynamic frequency-modulated signals; the dynamic frequency-modulated signal means that the frequency of the signal will change within a certain time range, and this method can be used to generate variable signals for testing and verifying the load response; the phase-locked loop is a control system that synchronizes the phase of the output signal with the phase of the input signal by adjusting the phase of the output signal, which is used in this method to track the fundamental wave of the target load by the frequency-modulated voltage signal, so as to obtain the phase information of the fundamental wave; the deep reinforcement learning is a branch of machine learning, which learns the optimal decision strategy through the interaction between the agent (Agent) and the environment, and in this method, the deep reinforcement learning is used to optimize the impedance and power factor of the load, and the feedback adjustment is performed through the frequency-modulated voltage signal; the load impedance is the comprehensive performance of the resistance and reactance of the load to the power supply, and in this test method, the load impedance is optimized by the deep reinforcement learning to achieve the optimal state of the load, so as to improve the efficiency and stability of the system; the power factor is an important indicator for measuring the power utilization efficiency in the power system, which represents the ratio of active power to apparent power; the wavelet transform is a signal analysis method that can decompose the signal into components with different frequency ranges, and the multi-scale wavelet transform refers to analyzing the signal through different scales (frequency bandwidths), which can capture both low-frequency and high-frequency information of the signal; the fundamental phase difference refers to the phase difference of the fundamental wave signal between different electrical systems or devices, and this parameter can help analyze the phase synchronization in the power system and judge whether the load is running stably; the harmonic refers to the frequency components other than the fundamental wave in the alternating current signal, and the harmonic energy distribution refers to the energy distribution of these non-fundamental wave frequency components, which can usually reflect the nonlinear characteristics of the load or the harmonic pollution in the power system.

[0017] In the second embodiment, the three-phase primary current and voltage analog load relay protection vector checking device test method is further provided with the following steps: generating a frequency modulation dynamic frequency modulation signal according to the fundamental wave parameters to obtain a frequency modulation voltage signal, including: A1, generating a frequency modulation index according to the Logistic chaotic mapping; A2, combining the frequency modulation index with the multi-harmonic injection to generate the frequency modulation voltage signal, wherein the calculation formula of the frequency modulation voltage signal is as follows: ; wherein, the frequency modulation voltage signal, the fundamental wave voltage, the fundamental wave frequency, the frequency modulation index, the harmonic random factor, simulating the harmonic fluctuation, the random initial phase of the harmonic, the angular frequency.

[0018] In this embodiment, the Logistic mapping is a simple mathematical model widely used to describe chaotic phenomena. It generates a series of numerical values through a recursive formula, and these values exhibit different dynamic behaviors from stability to chaos as the parameter changes. The frequency modulation index is an important parameter of the frequency modulation signal, which represents the ratio of the carrier frequency shift to the modulation signal frequency. In frequency modulation, the frequency of the signal changes according to the input modulation signal (such as speech, music, etc.), and the frequency modulation index describes the amplitude of this change. For example, when the frequency modulation index is large, the frequency change amplitude is also large. The fundamental wave voltage refers to the main frequency component (i.e. the lowest frequency) of the signal. In many waveforms, the fundamental wave is the strongest frequency. The fundamental wave frequency refers to the lowest frequency component in the signal, which determines the basic period of the signal. In alternating current, it is usually the frequency of the power grid. The harmonic random factor is used to simulate the harmonic fluctuation in the signal. Harmonics refer to frequency components that are integer multiples of the fundamental wave frequency. These frequency components often cause distortion in the waveform of the signal. The initial phase of the harmonic represents the starting phase of each harmonic component in the signal. Since harmonics have periodicity, the starting point (i.e. phase) of the waveform is an important parameter that determines the specific performance of the waveform.

[0019] In an optional embodiment, the fundamental wave of the target load is tracked according to the frequency modulation voltage signal through a phase-locked loop to obtain the fundamental wave phase, including: B1, obtaining the output phase of the phase-locked loop, determining the phase error according to the frequency modulation voltage signal and the output phase, and deriving the phase error to obtain the derivative of the phase error, wherein the calculation formula of the phase error is as follows: ; wherein, the phase error, This indicates the output phase of the phase-locked loop; B2. When the phase error is greater than the preset first error threshold and the derivative of the phase error is positive, the proportional gain of the phase-locked loop is increased linearly according to the preset growth weight. B3. When the phase error is less than the preset second error threshold and the derivative of the phase error is negative, the integral gain of the phase-locked loop is linearly reduced according to the preset weight reduction. B4. Track the fundamental wave of the target load using a phase-locked loop to obtain the fundamental wave phase.

[0020] It should be noted that the derivative of the phase error represents the rate at which the phase error changes with time, i.e., the trend of phase error change. By taking the derivative of the phase error, the speed of error change can be obtained, thus allowing for more effective adjustment of the phase-locked loop (PLL) control mechanism. The proportional gain is a control parameter in the PLL used to adjust the output signal. The proportional gain is proportional to the phase error, meaning that the larger the phase error, the larger the adjustment. The integral gain is another important control parameter in the PLL. It is related to the accumulated value of the phase error, i.e., the accumulation of the phase error over time.

[0021] In an optional embodiment, load impedance optimization of the target load is performed using a frequency-modulated voltage signal based on deep reinforcement learning to obtain the optimal load impedance and optimal power factor of the target load, including: C1. Define the state space, which includes the frequency modulation voltage and the current power factor of the target load; C2. Define the action space, which includes the resistance adjustment amount and the reactance adjustment amount; C3. Define the reward function, where the reward function is defined based on the adjustment amount of the current power factor, the target power factor, and the load impedance; C4. Measure the current voltage and power factor, and adjust the load impedance according to the current state; C5. By adjusting the resistance and reactance, the load impedance is updated to obtain a new voltage and a new power factor; C6. Calculate the reward based on the current state and the adjusted load impedance. The experience includes the state, action, reward and new state. Store the experience in the experience replay buffer. C7. Sample a batch of data from the experience replay buffer, update the Critic network by minimizing the error, update the Actor network using policy gradient ascent, and terminate the current round when the maximum number of steps is reached or the power factor stabilizes.

[0022] It should be noted that impedance is the ratio of current to voltage in a circuit, typically composed of resistance and reactance. Load impedance affects the power factor and current flow of the system. In reinforcement learning, the experience replay buffer is used to store the agent's historical experience (i.e., states, actions, rewards, and new states), which can then be used to train the model, particularly for learning from historical data. In reinforcement learning, the Critic network is used to evaluate the value of a state-action pair, i.e., the "value function." The Critic's task is to estimate the expected reward for a given state or state-action pair. The Actor network is another key component in reinforcement learning; it is responsible for generating the policy, i.e., deciding which action to choose in a given state. Policy gradient methods are a class of reinforcement learning algorithms that maximize cumulative rewards by directly optimizing the policy. Policy gradient ascent is the process of calculating gradients and updating the policy.

[0023] In an optional embodiment, the reward function is as follows: ;in, Represents the reward function, The penalty weight represents the power factor error. The penalty weight represents the magnitude of the load impedance adjustment. This indicates the adjustment amount of the load impedance. Indicates the current power factor. Indicates the target power factor; The updated formula for load impedance is as follows: ;in, Indicates the first Load impedance at the next iteration Indicates the resistance adjustment amount. This indicates the reactance adjustment amount.

[0024] In an optional embodiment, vector analysis is performed on the real-time voltage signal and real-time current signal based on multi-scale wavelet transform to obtain the fundamental phase difference and harmonic energy distribution of the target load, including: D1. Determine the scale parameters based on the sampling frequency and the actual frequency; D2. Determine the translation parameters based on the sampling frequency; D3. Determine the voltage wavelet coefficients corresponding to the real-time voltage signal and the current wavelet coefficients corresponding to the real-time current signal based on the scale parameters and translation parameters. D4. Determine the foundation dimensions based on the power frequency corresponding to the fundamental wave; D5. Extract the maximum energy ridge line at the fundamental scale based on the voltage wavelet coefficients; D6. Extract the first instantaneous phase of the real-time voltage signal and the second instantaneous phase of the real-time current signal at the maximum energy ridge. D7. Remove the basic scale, select other harmonic scales, and calculate the energy time-frequency distribution of each harmonic scale.

[0025] It should be noted that the fundamental scale is a scale related to the fundamental frequency (power frequency), which usually corresponds to the low-frequency components of the signal (such as 50Hz or 60Hz); the energy time-frequency distribution is a graph of the energy distribution of the signal in the time and frequency domains, used to analyze the time-frequency characteristics of the signal.

[0026] In an optional embodiment, the scale parameter is calculated using the following formula: ;in, Indicates the scale parameter. This represents the center frequency of the wavelet in the multi-scale wavelet transform. Indicates the sampling frequency. Indicates the actual frequency; The formula for calculating the translation parameters is as follows: ;in, Indicates the translation parameter; The formula for calculating the basic scale is as follows: ;in, Indicates the basic scale. Indicates the fundamental frequency; The formula for calculating the maximum energy ridge line is as follows: ;in, Indicates the ridge line of maximum energy. This represents the voltage wavelet coefficients.

[0027] In an optional embodiment, the formula for calculating the first instantaneous phase is as follows: ;in, Indicates the phase at the first instant; The formula for calculating the second instantaneous phase is as follows: ;in, Indicates the phase at the first instant; The formula for calculating the time-frequency distribution of energy is as follows: ;in, Represents the time-frequency distribution of energy. This indicates the total length of the real-time voltage signal.

[0028] In an optional embodiment, the protection verification of the target load is performed based on the fundamental phase difference and harmonic energy distribution to obtain the test label of the target load, including: E1, comparing the fundamental phase difference with a preset phase difference threshold, and comparing the harmonic energy distribution with a preset energy distribution threshold; E2. If the fundamental phase difference is greater than the preset phase difference threshold, or the harmonic energy distribution is greater than the preset energy distribution threshold, then the target load is judged to have harmonic interference; otherwise, the target load is judged not to have harmonic interference.

[0029] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A test method for a three-phase primary current and voltage simulated load-bearing relay protection vector inspection device, characterized in that, include: Obtain the fundamental wave parameters, and perform chaotic modulation of the dynamic frequency modulation signal based on the fundamental wave parameters to obtain the frequency-modulated voltage signal; The phase-locked loop tracks the fundamental wave of the target load using the frequency-modulated voltage signal to obtain the fundamental wave phase. Based on deep reinforcement learning, the target load is optimized using the frequency-modulated voltage signal to obtain the optimal load impedance and optimal power factor of the target load. Real-time signal acquisition is performed on the target load with optimal load impedance and optimal power factor to obtain real-time voltage and current signals. Vector analysis is performed on the real-time voltage signal and the real-time current signal using multi-scale wavelet transform to obtain the fundamental phase difference and harmonic energy distribution of the target load. The protection verification of the target load is performed based on the fundamental phase difference and the harmonic energy distribution to obtain the test label of the target load.

2. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 1, characterized in that, Based on the fundamental wave parameters, a chaotically modulated dynamic frequency-modulated signal is generated to obtain a frequency-modulated voltage signal, including: Generate the frequency modulation index based on the Logistic chaotic mapping; The frequency modulation index is combined with multiharmonic injection to generate a frequency-modulated voltage signal, wherein the calculation formula for the frequency-modulated voltage signal is as follows: ;in, Indicates a frequency-modulated voltage signal. Indicates the fundamental voltage. Indicates the fundamental frequency. Indicates the frequency modulation index. Represents the harmonic random factor, simulating harmonic fluctuations. This represents the random initial phase of the harmonic. It represents angular frequency.

3. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 2, characterized in that, The fundamental phase of the target load is obtained by tracking the fundamental frequency of the target load through the phase-locked loop using the frequency-modulated voltage signal, including: The output phase of the phase-locked loop is obtained, and the phase error is determined based on the frequency-modulated voltage signal and the output phase. The derivative of the phase error is then calculated, and the formula for calculating the phase error is as follows: ;in, Indicates phase error, This indicates the output phase of the phase-locked loop; When the phase error is greater than a preset first error threshold and the derivative of the phase error is positive, the proportional gain of the phase-locked loop is linearly increased according to a preset growth weight. When the phase error is less than a preset second error threshold and the derivative of the phase error is negative, the integral gain of the phase-locked loop is linearly reduced according to a preset reduction weight. The fundamental wave of the target load is tracked by the phase-locked loop to obtain the fundamental wave phase.

4. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 3, characterized in that, Based on deep reinforcement learning, the target load is optimized using the frequency-modulated voltage signal to obtain the optimal load impedance and optimal power factor of the target load, including: Define a state space, wherein the state space includes the frequency modulation voltage and the current power factor of the target load; Define an action space, wherein the action space includes resistance adjustment amount and reactance adjustment amount; Define a reward function, wherein the reward function is defined based on the adjustment amount of the current power factor, the target power factor, and the load impedance; Measure the current voltage and power factor, and adjust the load impedance accordingly; By adjusting the resistance and reactance, the load impedance is updated to obtain a new voltage and a new power factor; The reward is calculated based on the current state and the adjusted load impedance, wherein the experience includes state, action, reward and new state, and the experience is stored in the experience replay buffer; A batch of data is sampled from the experience replay buffer, the Critic network is updated by minimizing the error, the Actor network is updated using policy gradient ascent, and the current round is terminated when the maximum number of steps is reached or the power factor stabilizes.

5. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 4, characterized in that, The reward function is as follows: ;in, Represents the reward function, The penalty weight represents the power factor error. This represents the penalty weight for the load impedance adjustment magnitude. This indicates the adjustment amount of the load impedance. Indicates the current power factor. Indicates the target power factor; The update formula for the load impedance is as follows: ;in, Indicates the first Load impedance at the next iteration Indicates the resistance adjustment amount. This indicates the reactance adjustment amount.

6. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 5, characterized in that, Vector analysis is performed on the real-time voltage signal and the real-time current signal using multi-scale wavelet transform to obtain the fundamental phase difference and harmonic energy distribution of the target load, including: The scale parameters are determined based on the sampling frequency and the actual frequency. The translation parameters are determined based on the sampling frequency; The voltage wavelet coefficients corresponding to the real-time voltage signal and the current wavelet coefficients corresponding to the real-time current signal are determined based on the scale parameter and the translation parameter. The foundation dimensions are determined based on the power frequency corresponding to the fundamental wave. The maximum energy ridge at the fundamental scale is extracted based on the voltage wavelet coefficients. Extract the first instantaneous phase of the real-time voltage signal and the second instantaneous phase of the real-time current signal at the maximum energy ridge. Remove the base scale, select other harmonic scales, and calculate the energy time-frequency distribution of each harmonic scale.

7. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 6, characterized in that, The formula for calculating the scale parameter is as follows: ;in, Indicates the scale parameter. This represents the center frequency of the wavelet in the multi-scale wavelet transform. Indicates the sampling frequency. Indicates the actual frequency; The formula for calculating the translation parameters is as follows: ;in, Indicates the translation parameter; The formula for calculating the basic scale is as follows: ;in, Indicates the basic scale. The fundamental frequency is represented; the formula for calculating the maximum energy ridge is as follows: ;in, Indicates the ridge line of maximum energy. This represents the voltage wavelet coefficients.

8. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 7, characterized in that, The formula for calculating the first instantaneous phase is as follows: ;in, Indicates the phase at the first instant; The formula for calculating the second instantaneous phase is as follows: ;in, Indicates the phase at the first instant; The formula for calculating the time-frequency distribution of energy is as follows: ;in, Represents the time-frequency distribution of energy. This indicates the total length of the real-time voltage signal.

9. The test method for the three-phase primary current and voltage simulated load-bearing relay protection vector inspection device according to claim 8, characterized in that, The protection verification of the target load is performed based on the fundamental phase difference and the harmonic energy distribution to obtain the test label of the target load, including: The fundamental phase difference is compared with a preset phase difference threshold, and the harmonic energy distribution is compared with a preset energy distribution threshold. If the fundamental phase difference is greater than a preset phase difference threshold, or the harmonic energy distribution is greater than a preset energy distribution threshold, then the target load is determined to have harmonic interference; otherwise, the target load is determined not to have harmonic interference.

Images