Method, system and terminal for on-orbit evaluation of an electric thruster

Abstract

The present application belongs to the technical field of on-orbit application of spacecraft propulsion system, and relates to an on-orbit evaluation method and system for an electric thruster and a terminal. The on-orbit evaluation method comprises the following steps: obtaining multi-dimensional on-orbit telemetry data of the electric thruster during on-orbit operation, including electric system, thermal system, propellant delivery, and attitude control related parameters; using at least two methods of basic index calculation, neural network prediction, and data feature analysis to calculate the on-orbit performance, reliability, stability, and health status of the electric thruster; and fusing the calculation results to obtain corresponding evaluation results. Through the cooperation of multi-source data fusion and multi-dimensional algorithms, the present application realizes all-around accurate characterization of the equipment state, effectively improves the evaluation accuracy and robustness, and has engineering practical value.

Description

Technical Field

[0001] This invention relates to the field of on-orbit application technology of spacecraft propulsion systems, and in particular to an on-orbit evaluation method, system and terminal for electric thrusters. Background Technology

[0002] As the core actuator of a spacecraft propulsion system, the electric thruster, with its advantages of high specific impulse, long lifespan, and low propellant consumption, has become a key support for missions such as deep space exploration, satellite orbit maintenance, and attitude control. Its on-orbit performance, reliability, and stability directly determine the success or failure of spacecraft missions and their lifespan. Therefore, conducting accurate and comprehensive on-orbit evaluations is of great engineering significance.

[0003] However, existing electric thruster evaluation technologies suffer from significant shortcomings, including insufficient systematization, poor environmental adaptability, and weak data support, making it difficult to meet the high-precision evaluation requirements under complex on-orbit conditions. These shortcomings are specifically reflected in both the evaluation system and the basic data. Regarding evaluation systems, the three mainstream systems currently exist in several ways, all with significant technical shortcomings: First, there are ground-test-oriented performance calibration and aging evaluation systems. These systems calibrate performance parameters and predict lifespan by simulating vacuum and static environments at room temperature / high temperature on the ground. They rely heavily on ground static test data, but the fundamental differences between the space and ground environments mean that key conditions such as microgravity, space radiation, and dynamic mission loads cannot be accurately reproduced. This results in a deviation of 8% to 12% between performance predictions and actual on-orbit values, leading to significant deviations in lifespan predictions and making it difficult to support accurate on-orbit operation and maintenance decisions. Second, there are single-parameter threshold-type faults. The existing diagnostic system focuses on judging threshold exceedances of single performance or electrical parameters, neglecting the coupled effects of multiple dimensions such as electrical, thermal, attitude control, and propellant delivery parameters. This makes it unable to capture early, weak fault signals, resulting in delayed fault warnings and potentially causing on-orbit mission interruptions. Thirdly, the traditional telemetry data-based statistical condition monitoring system, while using onboard telemetry data, only collects a few core performance parameters and relies on basic statistical analysis algorithms. This results in a lack of data dimensions and insufficient analytical depth, failing to characterize the collaborative working characteristics of multiple systems in the electric thruster and the patterns of parameter degradation. The evaluation results are one-sided and cannot provide comprehensive support for operation and maintenance decisions. In summary, the existing evaluation system suffers from common problems such as low systematization, poor environmental adaptability, single evaluation dimensions, and insufficient data mining, failing to form a systematic evaluation capability adapted to complex dynamic on-orbit conditions.

[0004] Regarding basic data, the two types of basic data used in the existing evaluation cannot support accurate and comprehensive on-orbit evaluation: one is ground simulation test data, which is affected by the difference between the space and ground environments. The performance evaluation and life prediction based on this type of data deviate significantly from the actual on-orbit operating status and cannot truly reflect the dynamic working capability of the electric thruster on orbit; the other is traditional on-board telemetry data, which only collects a few core performance parameters and lacks multi-dimensional coupled data such as electrical, thermal, attitude control, and propellant delivery. The lack of data dimensions further exacerbates the one-sidedness of the evaluation results. Summary of the Invention

[0005] The purpose of this invention is to address the limitations of existing on-orbit evaluation technologies for electric thrusters, such as limited evaluation dimensions, poor environmental adaptability, low systematization, and insufficient data support. This invention proposes an on-orbit evaluation method, system, and terminal for electric thrusters. By enriching the dimensions of on-board telemetry data acquisition, integrating multi-source algorithms, and constructing a multi-dimensional evaluation system, this invention achieves a comprehensive and accurate characterization of the on-orbit status of electric thrusters, improves evaluation accuracy and robustness, and ensures the effective implementation of evaluations.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides an on-orbit evaluation method for an electric thruster, comprising the following steps: Acquire multi-dimensional on-board telemetry data of the electric thruster during its on-orbit operation. The multi-dimensional on-board telemetry data includes at least electrical system parameters, thermal system parameters, propellant delivery parameters, and attitude control related parameters. Based on multi-dimensional on-board telemetry data, at least two of the following methods are used to calculate the on-orbit performance, reliability, stability, and health status of the electric thruster: basic index calculation, neural network prediction, and data feature analysis. The calculation results are then fused to obtain on-orbit performance evaluation results, reliability evaluation results, stability evaluation results, and health status evaluation results.

[0007] As one possible approach, an overall on-orbit performance score is obtained by averaging the on-orbit basic performance score, performance change trend score, and degradation trend score. The on-orbit performance level is then determined based on the overall on-orbit performance score, thus obtaining the on-orbit performance evaluation result. The on-orbit basic performance score is calculated using basic indicators, specifically including the following steps: Based on the electrical system parameters including the grid output voltage / current, neutralizer holding output voltage / current, anode holding output voltage / current, anode output voltage / current, acceleration output voltage / current, and preset engineering parameters, calculate the timing sequence of thrust, specific impulse, and power. Calculate the mean, range, and standard deviation of thrust, specific impulse, and power; Based on the comparison of the mean, range, and standard deviation of thrust, specific impulse, and power with the engineering calibration parameters, the mean compliance score, range compliance score, and fluctuation stability score of thrust, specific impulse, and power are calculated respectively. The mean compliance score, range compliance score, and fluctuation stability score of thrust, specific impulse, and power are aggregated to obtain the single-parameter on-orbit performance score of thrust, specific impulse, and power. The weights of thrust, specific impulse, and power are determined using the entropy weight method. The on-orbit performance scores of individual parameters are weighted and summed to obtain the overall on-orbit performance score of the electric thruster.

[0008] As one possible implementation, both the performance change trend score and the degradation trend score are obtained using neural network prediction, specifically including the following steps: The time series of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output current / voltage, acceleration output / voltage, excitation power supply current, DC / AC voltage, and bus current, as well as thermal system parameters, including thruster temperature, and thruster delivery parameters, including low-pressure values, are input into a trained long short-term memory network model to obtain the anomaly probability value for each time step. Calculate the performance change trend score of the electric thruster based on the anomaly probability value; The electrical system parameters, including the grid output current / voltage, anode output current / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as the thermal system parameters, including the thruster temperature, are input into the trained deep autoencoder model to obtain the reconstruction error. The degradation trend score of the electric thruster is calculated based on the reconstruction error.

[0009] As one possible implementation method, the reliability evaluation results are obtained through the following approach: Acquire thrust, specific impulse, and power data during the ignition and operation period of the electric thruster; The thrust, specific impulse, and power data are logically compared with their respective safety threshold ranges, and the number of normal sampling points for each parameter during the ignition operation period is counted. The compliance rates of thrust, specific impulse, and power are calculated based on the ratio of the number of normal sampling points to the total number of sampling points during the ignition operation period. The compliance rates of thrust, specific impulse, and power are averaged and aggregated to obtain the overall compliance rate of performance parameters; Calculate the first-order difference sequence of thrust, specific impulse, and power during the operating period of the electric thruster; The absolute value of the first-order difference sequence is compared with a preset mutation threshold, and the mutation number of each parameter is counted. Calculate the single-parameter reliability of each parameter based on the number of mutations; The weights of thrust, specific impulse, and power are determined by using the entropy weight method. The reliability of each single parameter is then weighted and summed to obtain the on-orbit reliability of the electric thruster. The comprehensive reliability is obtained by merging the overall compliance rate of performance parameters and the on-orbit operational reliability with the arithmetic mean. The reliability level is then determined based on the comprehensive reliability, thus obtaining the reliability evaluation result.

[0010] As one possible implementation, the stability evaluation results are obtained through the following method: The thruster temperature, power processing unit temperature, thrust measurement unit temperature, and flow controller temperature from the thermal system parameters are obtained as a time series of temperature parameters, and the X-axis composite angular momentum, Y-axis composite angular momentum, and Z-axis composite angular momentum from the attitude control associated parameters are obtained as a time series of angular momentum parameters. Calculate the standard deviations of temperature and angular momentum parameters over multiple time windows, and calculate the single-parameter stability scores of the temperature control system and the attitude control system based on the ratio of the standard deviation of each parameter to the preset stability threshold. The stability scores of the temperature control system are weighted and summed to obtain the stability score of the temperature control system. The weights are determined by the entropy weight method based on the dispersion of each temperature parameter. The stability scores of the single parameters of the attitude control system are aggregated by mean, and the stability score of the attitude control system is obtained. The time sequence of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as thermal system parameters, including thruster temperature, and propellant delivery parameters, including low-pressure values, is input into the trained support vector machine model to obtain state recognition labels, and the support vector machine score is calculated based on the state recognition labels. The time sequence of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output current / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as thermal system parameters, including thruster temperature, and propellant delivery parameters, including low-pressure values, is input into the trained decision tree model to obtain state identification labels, and the decision tree score is calculated based on the state identification labels. The time series of parameters including the output current / voltage of the grid, the output current / voltage of the anode, the output current / voltage of the neutralizer heating, the acceleration output current, the excitation power supply current, and the bus current of the electrical system, the thruster temperature of the thermal system, and the low-pressure value of the propellant delivery parameters are input into the trained K-nearest neighbor model to obtain the matching similarity with the normal mode, and the K-nearest neighbor score is calculated based on the matching similarity. Calculate the coefficients of variation of thrust, specific impulse, and power, and calculate the fluctuation stability score of key performance parameters based on the coefficients of variation and the weights determined by the entropy weight method. The stability scores of the temperature control system, attitude control system, support vector machine, decision tree, K-nearest neighbor, and key performance parameter fluctuation stability are fused by arithmetic mean to obtain a comprehensive stability score. The stability level is then determined based on the comprehensive stability score, thus obtaining the stability evaluation result.

[0011] As one possible approach, health status assessment results are obtained through the following method: The electric thruster current health status is calculated based on the deviation of the current parameters from the rated current, including the grid output current, anode output current, anode holding output current, neutralizer holding output current, acceleration output current, excitation power supply current, and bus current. The temperature of the thruster, power processing unit, thrust measurement unit, and flow controller in the thermal system are obtained. The correlation coefficients between each pair of temperature parameters are calculated, and a normalized health correlation matrix is ​​constructed based on the correlation coefficients. The coupling health of multiple temperature parameters is calculated based on the mean of the off-diagonal elements of the normalized health correlation matrix. The time series of cathode heating / ablation power supply output current, anode output current, anode output voltage, grid output current, grid output voltage, acceleration output current, neutralizer holding output current, neutralizer holding output voltage, excitation power supply current, DC / AC voltage, thruster temperature, bus current, and low-pressure value from multi-dimensional on-board telemetry data are input into the trained XGBoost model to obtain anomaly probability values, and the XGBoost health score is calculated based on the anomaly probability values. The time series of the screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, thruster temperature, power processing unit temperature, thrust measurement unit temperature, and low-pressure value from the multi-dimensional on-board telemetry data are input into the trained local outlier model to obtain the local outlier score, and the local outlier health score is calculated based on the local outlier score. The time series of the screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, thruster temperature, power processing unit temperature, thrust measurement unit temperature, and low-pressure value from the multi-dimensional on-board telemetry data are input into the trained isolated forest model to obtain the anomaly probability value, and the isolated forest health score is calculated based on the anomaly probability value. The time series of screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, and thruster temperature from multi-dimensional on-board telemetry data are input into a trained one-dimensional convolutional neural network model to obtain the health state probability, and the health score of the one-dimensional convolutional neural network is calculated based on the health state probability. Extract the minimum value among the XGBoost health score, the local outlier health score, the isolated forest health score, and the one-dimensional convolutional neural network health score, and use it as the minimum health score of the neural network. The arithmetic mean of the electric thruster current health status, multi-temperature parameter coupled health status, and neural network minimum health score is fused to obtain a comprehensive health score. The health level is then determined based on the comprehensive health score, thus obtaining the health status evaluation result.

[0012] In a second aspect, the present invention provides an on-orbit evaluation system for electric thrust, comprising: The data acquisition module acquires multi-dimensional on-board telemetry data of the electric thruster during its on-orbit operation. The multi-dimensional on-board telemetry data includes at least electrical system parameters, thermal system parameters, propellant delivery parameters, and attitude control related parameters. The performance evaluation module is used to evaluate the on-orbit performance of the electric thruster based on multi-dimensional on-board telemetry data and obtain the performance evaluation results. The reliability evaluation module is used to evaluate the reliability of the electric thruster based on multi-dimensional on-board telemetry data and obtain the reliability evaluation results. The stability evaluation module is used to evaluate the stability of the electric thruster based on multi-dimensional on-board telemetry data and obtain the stability evaluation results. The health status assessment module is used to assess the health status of the electric thruster based on multi-dimensional on-board telemetry data and obtain the health status assessment results. Among them, at least one of the performance evaluation module, reliability evaluation module, stability evaluation module and health status evaluation module is specifically used to: perform calculations using at least two of the basic index calculation unit, neural network prediction unit and data feature analysis unit, and fuse the calculation results to obtain the evaluation result of the evaluation dimension; The output module is used to output performance evaluation results, reliability evaluation results, stability evaluation results, and health status evaluation results.

[0013] As one possible implementation, the performance evaluation module includes: The basic performance calculation unit is used to calculate the time series of thrust, specific impulse and power based on the grid output voltage, grid output current, neutralizer contact voltage, neutralizer contact current, anode contact voltage, anode contact current, anode voltage, anode current, acceleration voltage, acceleration current and preset engineering parameters from multi-dimensional on-board telemetry data. It also calculates the overall on-orbit performance score of the electric thruster based on the comparison of the mean, range and standard deviation of thrust, specific impulse and power with the engineering calibration parameters. The first neural network prediction unit is used to input the time series of the grid output current, grid output voltage, anode output current, anode output voltage, neutralizer heating output current, neutralizer heating output voltage, neutralizer holding output current, neutralizer holding output voltage, acceleration output current, acceleration output voltage, excitation power supply current, DC / AC voltage, thruster temperature, bus current, and low-pressure value from the multi-dimensional on-board telemetry data into the trained long short-term memory network model, and calculate the electric thruster performance change trend score based on the abnormal probability value output by the model. The second neural network prediction unit is used to input the time series of screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, and thruster temperature from multi-dimensional on-board telemetry data into the trained deep autoencoder model, and calculate the electric thruster degradation trend score based on the reconstruction error output by the model. The performance fusion unit is used to fuse the on-orbit comprehensive performance score, performance change trend score, and degradation trend score of the electric thruster by arithmetic mean to obtain the comprehensive performance score.

[0014] As one possible implementation, the reliability evaluation module includes: The basic performance calculation unit is used to acquire thrust, specific impulse and power data during the ignition working period of the electric thruster, and calculate the overall compliance rate of performance parameters by comparing the data with the safety threshold range. The data feature analysis unit is used to calculate the first-order difference sequence of thrust, specific impulse and power during the working period of the electric thruster. By comparing the absolute value of the first-order difference sequence with the mutation threshold, the on-orbit reliability of the electric thruster is calculated. The reliability fusion unit is used to fuse the comprehensive compliance rate of performance parameters and the on-orbit operational reliability by arithmetic mean to obtain the comprehensive reliability.

[0015] Thirdly, the present invention provides a terminal, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the on-orbit evaluation method for electric thrusters provided by the present invention.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention proposes an on-orbit evaluation method for electric thrusters, enriching the dimensions of on-board telemetry data acquisition and overcoming the problems of weak data support and poor environmental adaptability in existing systems. It expands the scope of telemetry data acquisition, comprehensively covering multiple key parameters such as electrical systems, thermal systems, propellant delivery, attitude control correlation, and mission performance. It encompasses core parameters including grid / anode / neutralizer current / voltage, excitation power supply current, component temperatures, low-pressure values, X / Y / Z-axis combined angular momentum, thrust, and specific impulse, providing sufficient and comprehensive data sources for multi-dimensional evaluation. This ensures that the evaluation results closely match the complex dynamic conditions under on-orbit conditions, improving the environmental adaptability and data support capabilities of the evaluation.

[0017] 2. This invention proposes an on-orbit evaluation method for electric thrusters, which integrates multi-source evaluation calculation methods to significantly improve evaluation accuracy and robustness. For each core evaluation dimension, three methods—basic index calculation, neural network prediction, and data feature analysis—are used in synergistic computation. Basic index calculation can directly quantify core parameters, the neural network model can deeply mine data time-series trends and hidden degradation patterns, and data feature analysis can accurately extract abnormal features. Through multi-source result fusion processing, the method effectively solves the problems of single and insufficient accuracy in existing evaluation algorithms, improving the reliability and anti-interference ability of evaluation results.

[0018] 3. This invention proposes an on-orbit evaluation method for electric thrusters, constructing a multi-dimensional evaluation system to effectively solve the technical problems of existing evaluations being limited in scope and providing only partial results. Focusing on the core requirements of on-orbit propulsion for electric thrusters, a multi-dimensional evaluation framework covering on-orbit performance, reliability, stability, and health status is established. This framework can comprehensively characterize the collaborative working state of the electric thruster's electrical, thermal, mechanical, and attitude systems, avoiding decision-making biases caused by single-dimensional evaluations and providing comprehensive and systematic evaluation support for on-orbit operation and maintenance decisions. Attached Figure Description

[0019] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of the on-orbit evaluation method for electric thrusters in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the process of obtaining on-orbit performance evaluation results of an electric thruster in an embodiment of the present invention. Figure 3 This is a flowchart illustrating the process of obtaining the reliability evaluation results of the electric thruster in an embodiment of the present invention. Figure 4 This is a flowchart illustrating the process of obtaining the stability evaluation results of the electric thruster in an embodiment of the present invention. Figure 5This is a flowchart illustrating the process of obtaining the health status evaluation results of an electric thruster in an embodiment of the present invention. Detailed Implementation

[0020] To facilitate a clear description of the technical solutions in the embodiments of the present invention, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. For example, the first threshold and the second threshold are merely used to distinguish different thresholds and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" are not necessarily different.

[0021] It should be noted that in this invention, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0022] In this invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one" or similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, "at least one of a, b, or c" can represent: a, b, c, a combination of a and b, a combination of a and c, a combination of b and c, or a, b, and c, where a, b, and c can be single or multiple.

[0023] In a first aspect, embodiments of the present invention provide an on-orbit evaluation method for an electric thruster, see [link to relevant documentation]. Figure 1 It includes the following steps: (i) Acquiring multi-dimensional on-board telemetry data Acquire multi-dimensional on-board telemetry data of the electric thruster during its on-orbit operation. The multi-dimensional on-board telemetry data includes at least electrical system parameters, thermal system parameters, propellant delivery parameters, and attitude control related parameters. Among them, electric thrusters are the core actuators of spacecraft propulsion systems. With their advantages of high specific impulse, long lifespan, and low propellant consumption, they provide key support for missions such as deep space exploration, satellite orbit maintenance, and attitude control. They generate thrust by ionizing electrical energy and accelerating propellant. Their performance and reliability depend on precise monitoring of their operational status, and their on-orbit status directly affects the success or failure of spacecraft missions and their service life.

[0024] The electrical system parameters include grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output current / voltage, acceleration output / voltage, excitation power supply current, DC / AC voltage, bus current, anode holding output voltage / current, and cathode heating / ablation power supply output voltage / current. 1) The output current of the grid is a core parameter of the grid's operating status, directly reflecting its load condition. Abnormalities may indicate grid failure or uneven plasma distribution, threatening thrust stability. This parameter is a continuous numerical time-domain data. Under normal conditions, it is stably positively correlated with thrust, with small start-up overshoot and fluctuation amplitude, and a smooth waveform. Abnormalities result in sudden current rises and falls, failure of the thrust correlation, and extreme values ​​exceeding the safe range. Continuous abnormalities will cause the thrust to deviate from mission requirements, necessitating investigation of residual grid working fluid, insulation damage, or structural deformation.

[0025] 2) The output voltage of the grid is a key parameter to ensure normal arc initiation of the grid. Its stability directly affects the beam acceleration efficiency. Fluctuations can lead to abnormal beam energy and reduced propulsion efficiency. This parameter is a continuous numerical time-domain data. Under normal conditions, it is stably positively correlated with the specific impulse, with small steady-state fluctuations, uniform data distribution, and small ripple. Under abnormal conditions, the voltage rises and falls sharply, the correlation with the specific impulse breaks, and fluctuations and ripples intensify. This not only reduces propulsion efficiency but may also cause plasma oscillations, requiring adjustment of power supply parameters or repair of the grid structure.

[0026] 3) The anode output current is the core parameter for monitoring the anode's operating status. Its changes reflect the anode load and discharge chamber conditions. Abnormalities may indicate discharge chamber malfunctions, affecting the continuity of thrust output. This parameter is a continuous numerical time-domain data. Under normal conditions, it is stably positively correlated with the working fluid flow rate, with small fluctuations, concentrated data, and regular waveforms without abrupt changes. When abnormal, the current drops sharply or high-frequency arcing jumps occur, the correlation with flow rate fails, extreme values ​​easily exceed the design limit, and electromagnetic interference may also affect the control circuit. It is necessary to check the working fluid supply or the magnetic field status of the discharge chamber.

[0027] 4) The anode output voltage is a fundamental parameter for the normal operation of the anode. Its stable supply directly ensures the generation of plasma in the discharge chamber. Abnormalities can lead to unstable discharge or even interruption of thrust output. This parameter is a continuous numerical time-domain data. Under normal conditions, the steady-state fluctuation is small and the waveform is smooth, matching the anode current. Under abnormal conditions, the voltage rises and falls sharply, the fluctuation intensifies, the value becomes unstable, and the matching relationship with the anode current is broken. Long-term abnormalities will accelerate anode wear and shorten its lifespan, requiring maintenance of the discharge circuit or anode insulation structure.

[0028] 5) The neutralizer heating output current is the core parameter for neutralizer activation. Its stability directly ensures the heating effect. Abnormalities will cause the neutralizer to fail to start normally and affect the charge neutralization function. This parameter is a continuous numerical time-domain data. Under normal conditions, it changes according to the startup sequence, with small fluctuations and is coordinated with the neutralizer temperature. In case of abnormalities, the current is too small, which will cause ignition failure, and too large, which will accelerate cathode wear. It is necessary to adjust the heating power supply parameters or repair the temperature control logic.

[0029] 6) The neutralizer heating output voltage is a fundamental parameter of the neutralizer heating system. Its voltage value directly affects the heating efficiency, and deviations can lead to insufficient heating or start-up failures. This parameter is a continuous numerical time-domain data in a low voltage range. Under normal conditions, it is stable within the design range with minimal fluctuations and is coordinated with the output current. In abnormal conditions, the voltage rises or falls sharply, the coordination relationship fails, resulting in abnormal heating power, requiring inspection of the heating power supply or the neutralizer circuit.

[0030] 7) The neutralizer's contact output current is a key parameter of the neutralizer's contact mechanism. Its stable output ensures the neutralizer's normal operation, while abnormalities lead to contact failure and reduced charge neutralization effect. This parameter is a continuous time-domain data in the microampere to milliampere range. Under normal conditions, the value is small, fluctuating smoothly and regularly before and after ignition. When abnormal, it remains consistently high and does not match the ignition state, reflecting cathode activity decay or circuit failure, which will affect the neutralizer's electron emission efficiency and cause the beam to fail to neutralize effectively.

[0031] 8) The neutralizer contact output voltage is a fundamental parameter of the neutralizer contact system. Its stability directly determines the contact reliability, and fluctuations can easily lead to malfunctions and charge accumulation in the spacecraft. This parameter is continuous time-domain data. Under normal conditions, it is stable within the design range, with small fluctuations and is matched with the contact current. In abnormal conditions, the voltage rises and falls sharply, the matching relationship fails, causing abnormal neutralizer potential, reducing ignition success rate and electron emission stability, requiring inspection of the contact power supply or neutralizer structure.

[0032] 9) The accelerating output current is a core parameter in the plasma acceleration process, and its magnitude directly determines the beam acceleration effect. Abnormalities will lead to insufficient beam velocity and reduced thrust specific impulse. This parameter is a continuous numerical time-domain data in the microampere to milliampere range. Under normal conditions, the value is extremely small, with minimal fluctuations and does not interfere with the accelerating voltage. When abnormal, the current increases sharply, indicating that the accelerating gap is breaking down. If the current remains high, it will cause the accelerating grid to overheat and be damaged, affecting the normal operation of the grid and anode. It is necessary to immediately shut down the accelerating power supply and check the gap insulation status.

[0033] 10) The acceleration output voltage is a key parameter to ensure effective beam acceleration. Its stability directly affects propulsion efficiency, and fluctuations can lead to uneven beam acceleration and thrust fluctuations. This parameter is a continuous numerical time-domain data. Under normal conditions, it is stably correlated with ion acceleration efficiency, with small fluctuations and minimal ripple. When abnormal, the voltage rises and falls sharply, the correlation with acceleration efficiency breaks, and fluctuations intensify. This not only reduces propulsion efficiency but may also cause ion beam divergence and affect thrust pointing accuracy. Adjustments to the acceleration power supply or repairs to the acceleration gap are necessary.

[0034] 11) Excitation Power Supply Current: The excitation power supply current is a key parameter for generating the confined plasma magnetic field. Its magnitude directly determines the magnetic field strength. Abnormalities can lead to plasma confinement failure, beam divergence, and reduced propulsion efficiency. This parameter is a continuous numerical time-domain data. Under normal conditions, it is stably correlated with the magnetic field strength, with small fluctuations and a smooth waveform. Abnormalities can cause a sudden increase or decrease in current, leading to magnetic field confinement failure and a decrease in thrust. Continued abnormalities can damage the excitation coil or discharge chamber, requiring emergency shutdown and maintenance.

[0035] 12) DC / AC voltage is a fundamental parameter for the power supply of various systems in the thruster. Its stability ensures the overall reliability of the power supply, while fluctuations can cause abnormal operation of various components and threaten the normal operation of the thruster. This parameter is a continuous numerical time-domain data. Under normal conditions, the input and output voltages are stable, with small fluctuations and low ripple, which is suitable for the needs of downstream modules. Under abnormal conditions, the ripple increases and the conversion efficiency decreases, causing abnormal operation of downstream modules. It is necessary to check the filter circuit or conversion logic.

[0036] 13) Bus current is a core indicator of the load status of the power supply system. Its changes reflect the overall load situation, and abnormalities may indicate component failures or lead to power overload. This parameter is a continuous numerical time-domain data. Under normal conditions, it changes with the start-up and shutdown patterns of the modules, and in steady state, it satisfies current conservation with small fluctuations. When abnormal, the current suddenly increases or decreases, and the conservation relationship fails. Downstream modules need to be checked to avoid overload or power outage shutdowns.

[0037] The thermal system parameters include thruster temperature, power processing unit temperature, thrust measurement unit temperature, and flow controller temperature. 14) Thruster temperature is a key parameter for monitoring its operating status and directly affects the reliability of components. Excessive temperature accelerates material aging and damages components, while insufficient temperature affects propellant vaporization. This parameter is a continuous numerical time-domain data. Under normal conditions, the temperature of key components is within the design range with small fluctuations and a reasonable temperature gradient between components. In case of abnormal conditions, the temperature rises or falls sharply and exceeds the safety threshold, requiring power adjustment or overhaul of the heat dissipation system.

[0038] 15) The temperature of the power processing unit is a core indicator of its operating status. Its stability ensures power conversion efficiency, while excessively high temperatures can lead to performance degradation and trigger protection shutdown. This parameter is a continuous numerical time-domain data. Under normal conditions, the component temperature is within the heat dissipation adaptation range with small fluctuations and reasonable temperature differences between components. In abnormal conditions, the temperature rises rapidly, triggering power reduction or shutdown, requiring inspection of the heat dissipation system or optimization of power allocation.

[0039] 16) The temperature of the thrust measurement unit is a key parameter of the thrust measurement unit. Its stability ensures the accuracy of thrust measurement, while abnormalities can lead to measurement deviations and affect orbit control decisions. This parameter is a continuous numerical time-domain data. Under normal conditions, the temperature is close to room temperature, with small fluctuations and uniform distribution. In case of abnormalities, overheating will cause a decrease in the accuracy of the control chip, requiring optimization of the installation position or repair of the internal circuitry.

[0040] 17) The temperature of the flow controller is the core parameter for propellant flow control. Its stability ensures the accuracy of flow regulation, while abnormalities can lead to flow control failure and uneven propellant supply. This parameter is a continuous numerical time-domain data. Under normal conditions, the temperature is within the stable range of the working fluid with small fluctuations; under abnormal conditions, the temperature is too low, causing the working fluid to condense and the flow rate to drop sharply, requiring inspection of the heating element or insulation structure.

[0041] 18) The anode contact output current is a key parameter of the anode contact circuit. Its stability directly affects the reliability of the anode operation. Abnormalities can lead to poor anode contact and discharge faults. This parameter is a continuous numerical time-domain data in the microamp to milliamp range. Under normal conditions, the value is close to zero, with small fluctuations and no interference with the anode output voltage. When abnormal, the current increases sharply, indicating damage to the anode insulation structure. A sustained high current can cause an imbalance in the potential between the anode and the casing, affecting plasma generation and even causing a short circuit, requiring emergency shutdown and repair.

[0042] 19) The anode contact output voltage is the core parameter of the anode contact system. Its voltage value directly ensures the effectiveness of the contact function. Deviation will lead to contact failure and threaten the normal operation of the anode. This parameter is a continuous numerical time-domain data. Under normal conditions, it is stable within the design range, with small fluctuations and no obvious correlation with the contact current. Under abnormal conditions, the voltage rises and falls sharply, the matching relationship with the current is broken, and continuous abnormality will damage the anode insulation and increase the risk of leakage current. It is necessary to check the contact power supply or the anode insulation status.

[0043] 20) The output voltage of the cathode heating / ablation power supply is the core parameter for cathode activation and maintenance. Its stability directly ensures the normal operation of the cathode. Abnormalities can lead to insufficient cathode heating or excessive ablation, shortening its lifespan. This parameter is a continuous numerical time-domain data in a low voltage range. Under normal conditions, it is stable within the design range with small fluctuations and matches the output current. Under abnormal conditions, the voltage rises and falls sharply, the matching relationship fails, resulting in abnormal heating power, requiring repair of the heating power supply or cathode circuit.

[0044] 21) The output current of the cathode heating / ablation power supply is a key parameter of the cathode heating system. Its magnitude directly controls the heating power. Abnormalities can lead to cathode activation failure or excessively rapid ablation, affecting the thruster's startup and operation. This parameter is a continuous numerical time-domain data. Under normal conditions, it varies with the startup sequence, fluctuates little, and is correlated with the cathode temperature. In abnormal conditions, the current remains excessively high or zero. Excessive current accelerates cathode ablation, while zero current prevents cathode activation. The circuit or temperature sensor needs to be checked.

[0045] Among them, the propellant delivery parameters include the low-pressure value; 22) Low-pressure value is a key indicator of the propellant delivery system. Its stability directly ensures the continuous supply of propellant. Abnormalities can lead to insufficient or excessive propellant supply, affecting the stability of thrust output. This parameter is a continuous numerical time-domain data. Under normal conditions, the monitoring area is in a high-vacuum design state with small fluctuations, and the pressure is stable near the lower limit of the design. When abnormal, the pressure rises or falls sharply, deviating from the vacuum range. A sharp rise reduces ionization efficiency, while a sharp drop reflects a working fluid supply failure, requiring investigation of pipeline sealing or the condition of the working fluid storage tank.

[0046] Among them, the attitude control associated parameters include the X-axis composite angular momentum, the Y-axis composite angular momentum, and the Z-axis composite angular momentum; 23) The combined angular momentum of the X / Y / Z axes is a core parameter reflecting the spacecraft's attitude control capability. Its magnitude and direction directly affect the attitude adjustment accuracy. Anomalies can lead to attitude drift and threaten the stable operation of the spacecraft. This parameter is a continuous numerical time-domain data. Under normal conditions, it works in conjunction with the attitude control system, is numerically stable, and has a reasonable distribution across axes. In anomalies, the angular momentum of a single axis increases significantly, causing attitude deviation. If this deviation exceeds the correction capability, it will trigger attitude oscillations, requiring a reduction in thrust power or adjustment of thrust direction, and activation of the attitude correction device.

[0047] (II) Multi-dimensional evaluation Based on multi-dimensional on-board telemetry data, at least two of the following methods are used to calculate the on-orbit performance, reliability, stability, and health status of the electric thruster: basic index calculation, neural network prediction, and data feature analysis. The calculation results are then fused to obtain on-orbit performance evaluation results, reliability evaluation results, stability evaluation results, and health status evaluation results.

[0048] 1. On-orbit performance As one possible implementation, see Figure 2 The on-orbit comprehensive performance score is obtained by averaging the on-orbit basic performance score, performance change trend score, and degradation trend score. The on-orbit performance level is determined based on the on-orbit comprehensive performance score, thus obtaining the on-orbit performance evaluation result. As an example, let the overall on-orbit performance score be P. Then P = (on-orbit basic performance score + performance change trend score + degradation trend score) ÷ 3, and the result is rounded to two decimal places. Grading rules: 9 ≤ P ≤ 10 points, excellent; 7 ≤ P < 9 points, good; 6 ≤ P < 7 points, passing grade; P < 6 points, unqualified.

[0049] In summary, the on-orbit comprehensive performance score, calculated using basic indicators, can accurately quantify the compliance and stability of thrust, specific impulse, and power. The performance change trend score, using LSTM, excels at capturing long-term performance dependencies, while the degradation trend score, using DAE, can identify potential degradation patterns in an unsupervised manner. These three methods comprehensively cover performance evaluation needs from current, long-term, and implicit perspectives. Individual evaluations of each have limitations and require synergy and complementarity. By integrating the advantages of accurate quantification of basic indicators with the trend prediction and unsupervised detection advantages of neural networks, the one-sidedness of single-algorithm evaluation can be avoided. No preset weights are required, and the overall level and evolution trend of the electric thruster's on-orbit performance can be objectively reflected.

[0050] (1) On-orbit basic performance The on-orbit basic performance score is calculated using basic indicators, specifically including the following steps: Based on the electrical system parameters including the grid output voltage / current, neutralizer holding output voltage / current, anode holding output voltage / current, anode output voltage / current, acceleration output voltage / current, and preset engineering parameters, calculate the timing sequence of thrust, specific impulse, and power. The preset engineering parameters include: thrust compensation coefficient, k=0.95; xenon atomic mass, m=2.18×10−25kg; charge constant, e=1.6×10−19C; gravitational acceleration, g=9.8m / s2; and propellant flow rates in each branch: anode flow rate 1.09mg / s, main cathode flow rate 0.136mg / s, and neutralizer flow rate 0.136mg / s. Among these steps, the following steps are taken to collect electrical system parameters: the sampling frequency of the raw electrical parameters is synchronized with the telemetry downlink frequency, for example, the sampling frequency is 1~10Hz, to ensure complete capture of the electrical system status; lost telemetry data and abnormal jump values ​​without cause are removed, such as values ​​whose single jump amplitude exceeds the rated range by 20% without any change in external operating conditions; the proportion of valid sampling points within the data acquisition period is ≥95%, to avoid performance evaluation deviations caused by missing data; and all electrical parameter timestamps are fully aligned, for example, with an error ≤100ms, to ensure the accuracy of core performance parameter calculations.

[0051] As an example, the time series of thrust, specific impulse, and power are calculated: Thrust calculation: Where k is the thrust compensation coefficient, with a measured value of 0.95; m is the atomic mass of xenon, which is 2.18 × 10⁻²⁵ kg; and e is the charge constant, which is 1.6 × 10⁻¹⁹ C. The output voltage of the grid is expressed in volts (V). This represents the output current of the grid, measured in amperes (A).

[0052] Specific impulse calculation: in, The total propellant flow rate is the sum of the flow rates from the anode, cathode, and neutralizer, expressed in kg / s; g is the acceleration due to gravity, taken as 9.8 m / s². 2 The exact value of the on-orbit flow rate could not be determined, so based on ground calibration, the anode flow rate was 1.09 mg / s, the main cathode flow rate was 0.136 mg / s, and the neutralizer flow rate was 0.136 mg / s.

[0053] Power value calculation: Where P is the power of a single unit. For voltage, It represents electric current.

[0054] Calculate the mean, range, and standard deviation of thrust, specific impulse, and power; Among these, the mean, range, and standard deviation of thrust, specific impulse, and power are key indicators characterizing the overall on-orbit performance of a single electric thruster, and can be calculated from its original electrical parameters. The mean reflects the overall operational level, the range reflects the amplitude of extreme value fluctuations, and the standard deviation characterizes the degree of dispersion and parameter stability. The combination of these three parameters can comprehensively evaluate the on-orbit stability and operational compliance of the electric thruster. Insufficient performance parameters will directly reduce propulsion output accuracy and increase the risk of deviations in spacecraft orbit and attitude control.

[0055] Based on the comparison of the mean, range, and standard deviation of thrust, specific impulse, and power with the engineering calibration parameters, the mean compliance score, range compliance score, and fluctuation stability score of thrust, specific impulse, and power are calculated respectively. Among them, the engineering calibration parameters include: the rated range of thrust, specific impulse, and power, the maximum allowable range, and the maximum allowable standard deviation. The calibration parameters are determined based on the electric thruster design specifications and ground tests, and are the benchmark boundaries for performance scoring of core performance parameters.

[0056] As an example, the mean compliance score, range compliance score, and volatility stability score are all out of a total of 10 points, and the scoring rules are consistent. Average compliance score (out of 4) The mean is completely within the specified range → 4 points; If the mean value exceeds the rated range but deviates from the rated center value by ≤5%, then 2 points are awarded. If the mean exceeds the rated range and deviates from the rated center value by more than 5%, the score will be 0.

[0057] Extremely poor compliance score (out of 3) Range ≤ Maximum Allowable Range → 3 points; Maximum allowable range < range ≤ 1.2 × maximum allowable range → 1 point; Range > 1.2 × maximum allowable range → 0 points.

[0058] Fluctuation stability score (out of 3) Standard deviation ≤ maximum allowable standard deviation → 3 points; Maximum allowable standard deviation < standard deviation ≤ 1.2 × maximum allowable standard deviation → 1 point; Standard deviation > 1.2 × maximum allowable standard deviation → 0 points.

[0059] The mean compliance score, range compliance score, and fluctuation stability score of thrust, specific impulse, and power are aggregated to obtain the single-parameter on-orbit performance score of thrust, specific impulse, and power. As an example, the single-parameter on-orbit performance score = mean compliance score + range compliance score + fluctuation stability score.

[0060] The weights of thrust, specific impulse, and power are determined by using the entropy weight method. The on-orbit performance scores of the individual parameters are weighted and summed to obtain the basic on-orbit performance score of the electric thruster.

[0061] Among them, the entropy weight method determines the weights of thrust, specific impulse and power, that is, the distribution based on the degree of dispersion of the scores of the three types of parameters. The higher the degree of dispersion, the higher the weight ratio, and the more significant the impact on the overall performance.

[0062] In summary, the on-orbit basic performance evaluation calculates key indicators such as thrust, specific impulse, and power from the original electrical parameters, realizing the correlation mapping between the electrical system and propulsion performance, which can effectively reflect the cooperative matching characteristics of the two systems. At the same time, it conducts multi-dimensional scoring of core parameters based on mean, range, and standard deviation, comprehensively covering operational level and fluctuation characteristics, accurately identifying performance shortcomings, and significantly improving the completeness and pertinence of the evaluation.

[0063] As one possible implementation, both the performance change trend score and the degradation trend score are obtained using neural network prediction, specifically including the following steps: (2) Performance change trend The time series of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output current / voltage, acceleration output / voltage, excitation power supply current, DC / AC voltage, and bus current, as well as thermal system parameters, including thruster temperature, and thruster delivery parameters, including low-pressure values, are input into a trained long short-term memory network model to obtain the anomaly probability value for each time step. Among them, Long Short-Term Memory (LSTM) network, as an improved algorithm of Recurrent Neural Network (RNN), can effectively capture long-term dependencies in time-series data and is suitable for dynamic evaluation of the performance change trend of electric thrusters. LSTM selectively memorizes and forgets time-series information through a gating mechanism consisting of input gates, forget gates, and output gates. It can accurately learn the time-series evolution law of thruster parameters and output the anomaly probability at each time step, thereby realizing the effective judgment of its performance change trend.

[0064] The abnormal probability value ranges from 0 to 1. The closer the probability value is to 1, the higher the probability that the electric thruster is in an abnormal state at the current moment; the closer it is to 0, the closer the operating state is to normal.

[0065] Calculate the performance change trend score of the electric thruster based on the anomaly probability value; As an example, calculate the performance change trend score of the electric thruster: First, calculate the first difference of the outlier probability values ​​and count the total number of first differences; Summing the first-order differences, recording the positive and negative values ​​of the results; if the result is 0, directly output the performance change trend score = 0. In the first-order difference sequence, count the number of differences that have the same sign as the result in the previous step, and the number of differences that are different. Performance change trend score = (Number of identical differences - Number of different differences) / Total number × 10; The score quantifies the state evolution trend of the electric thruster on a 10-point scale. The higher the score, the more significant the trend of the state deteriorating, providing a dynamic basis for early warning of potential failures.

[0066] (3) Deterioration trend The electrical system parameters, including the grid output current / voltage, anode output current / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as the thermal system parameters, including the thruster temperature, are input into the trained deep autoencoder model to obtain the reconstruction error. Among them, the deep autoencoder (DAE) is an unsupervised deep learning algorithm that can automatically extract latent features from high-dimensional data, making it suitable for unsupervised evaluation of the performance degradation trend of electric thrusters. This algorithm maps input data to a low-dimensional feature space through an encoder, and then the decoder reconstructs the data. The reconstruction error characterizes the degree of deviation between the data and the normal pattern, thereby enabling the judgment of performance degradation trends and meeting the needs of monitoring complex features of thrusters.

[0067] The higher the reconstruction error value, the more the current operating state of the electric thruster deviates from the reconstruction result of the normal mode; the lower the error value, the closer the state is to normal.

[0068] The degradation trend score of the electric thruster is calculated based on the reconstruction error.

[0069] As an example, calculate the degradation trend score of the electric thruster: First, calculate the temporal rate of change of the reconstruction error, i.e., the first difference; a positive rate of change indicates that the error is increasing and the degradation is aggravated, while a negative rate of change indicates that the error is decreasing and the state is improving. Furthermore, the sliding window statistics show the proportion of days with a negative rate of change, i.e., the proportion of days with improved conditions. The health trend index = the proportion of negative change rate within the sliding window × (1 - normalized value of average reconstruction error max / min); where the average reconstruction error is mapped to [0,1] using the max / min method, the lower the error, the closer the normalized value is to 0, and the closer (1 - normalized value) is to 1. Deterioration trend score = health trend index × 10; the score range is mapped to [0,10]; if the health trend index has been normalized, multiply directly; if it has not been normalized, first perform max / min normalization on the index and then multiply by 10.

[0070] A higher degradation trend score indicates a smaller reconstruction error, a higher degree of state improvement, and a more significant healthy development trend; a lower score indicates a larger reconstruction error, a significant degradation trend, and a potential risk of performance degradation, which can provide dynamic support for early warning and operation and maintenance decisions in orbit.

[0071] 2. Reliability As one possible implementation, see Figure 3 The reliability evaluation results were obtained using the following methods: (1) Overall compliance rate of performance parameters Acquire thrust, specific impulse, and power data during the ignition and operation period of the electric thruster; Among them, thrust reflects the stability of propulsion power output and directly determines the accuracy of orbit adjustment; specific impulse reflects the level of propulsion energy efficiency and is related to the spacecraft's mission endurance; power reflects energy consumption and conversion efficiency and is related to the coordinated stability of the power supply system.

[0072] Specifically, the data acquisition period and open parameters must be strictly limited to the electric thruster ignition operating period, excluding data from ignition start-up transients, shutdown braking transients, and non-ignition periods. Data is based on the electric thruster ignition operating period, and invalid data from non-ignition periods is excluded.

[0073] The thrust, specific impulse, and power data are logically compared with their respective safety threshold ranges, and the number of normal sampling points for each parameter during the ignition operation period is counted. Among them, each safety threshold range is based on the electric thruster design specifications and ignition working condition calibration, distinguishing the thresholds corresponding to different ignition power levels.

[0074] As an example, perform a two-branch logic operation on the real-time data of each sampling point during the ignition operation period of the electric thruster: If the lower threshold ≤ parameter ≤ upper threshold, it is determined that the state is normal and the state quantity is recorded as 0. If the parameter is less than the lower threshold or the parameter is greater than the upper threshold, it is determined that the state is out of limit and the state variable is recorded as 1. Attached over-limit information: over-limit type and over-limit core data; over-limit types include: insufficient thrust over-limit, excessive specific impulse over-limit, and power overload over-limit; over-limit core data includes: current parameter value, over-limit start time, and duration, where the over-limit start time is accurate to the second, limiting the ignition working period.

[0075] Unified identification of exceeding limits provides rapid early warning of malfunctions during ignition operation.

[0076] As an example, add debouncing logic: Only when three consecutive sampling points are all determined to be 1 is it finally confirmed that the parameter is valid due to exceeding the limit; if the parameter returns to normal immediately after exceeding the limit in a single frame, it is determined to be an instantaneous interference, and the correction state quantity is 0, so as to ensure the accuracy of the warning under the ignition condition and avoid misjudgment caused by instantaneous plasma fluctuations and telemetry noise during ignition.

[0077] The compliance rates of thrust, specific impulse, and power are calculated based on the ratio of the number of normal sampling points to the total number of sampling points during the ignition operation period. As an example, for the valid sampling points during the ignition operation period, the number of times the state variable = 0 is counted. The results are rounded to two decimal places. If the number of valid sampling points during the ignition period is insufficient, the data must be marked as invalid and will not be included in the compliance rate calculation.

[0078] The compliance rates of thrust, specific impulse, and power are averaged and aggregated to obtain the overall compliance rate of performance parameters; Among them, the overall compliance rate of performance parameters is the core quantitative indicator of the reliable operation level of electric thrusters under ignition conditions. The higher the score, the closer the thrust, specific impulse, and power parameters are to the safety threshold requirements, the more stable and reliable the propulsion system is, and the more guaranteed the quality of core tasks such as spacecraft orbit adjustment and attitude control are.

[0079] (2) On-orbit operational reliability The time series of thrust, specific impulse and power during the working period of the electric thruster are collected, and data from non-working periods, such as shutdown, ignition start-up transients and shutdown braking transients, are strictly excluded to ensure the accuracy of the calculations. The time sequence must be the real-time value transmitted from on-orbit telemetry, with perfectly aligned timestamps. If the error is ≤100ms, the accuracy of the differential calculation must be ensured.

[0080] Calculate the first-order difference sequence of thrust, specific impulse, and power during the operating period of the electric thruster; Among them, the first-order difference algorithm can effectively capture the characteristics of sudden data changes by calculating the difference between adjacent data points, which is suitable for the on-orbit parameter monitoring requirements of electric thrusters. This algorithm can accurately identify sudden abnormal changes in thrust, specific impulse, and power, providing a basis for judging problems such as abnormal plasma generation, propellant delivery failure, and power supply mismatch, and realizing early warning of faults.

[0081] As an example, first-order difference computation and difference sequence generation; First-order difference = next data point - previous data point By traversing the time-domain data of the electric thruster's operating parameters, eliminating the overall trend and highlighting local abrupt changes, calculating the difference between adjacent data points, and constructing first-order difference sequences for thrust, specific impulse, and power respectively, the resulting difference fluctuation curves visually demonstrate the characteristics of parameter abrupt changes. The absolute value of the first-order difference sequence is compared with a preset mutation threshold, and the mutation number of each parameter is counted. Among them, the setting of individual mutation thresholds for the three parameters of thrust, specific impulse and power should be based on the operating conditions and design specifications of the electric thruster, and the mutation thresholds corresponding to different operating power levels should be distinguished to determine whether the differential results belong to sudden anomalies.

[0082] Each parameter is determined to be either mutated or not, the total number of mutations within the selected period is counted, and the time of mutation occurrence is recorded with precision down to the second.

[0083] As an example, for the difference sequences of each parameter, when the absolute value of the difference exceeds the preset mutation threshold, it is determined to be an abnormal mutation. The total number of mutations of the parameters within the set period is counted, and the number of normal mutations is obtained by subtracting the number of mutations from the number of valid data points in the time period, providing a basis for single-parameter reliability calculation.

[0084] Calculate the single-parameter reliability of each parameter based on the number of mutations; As an example, single-parameter reliability is denoted as... The expression is: = (Number of normal attempts / Total number of attempts) × 100% The weights of thrust, specific impulse, and power are determined by using the entropy weight method. The reliability of each single parameter is then weighted and summed to obtain the on-orbit reliability of the electric thruster. Among them, based on the dispersion of the difference sequences of each parameter, the entropy weight method is used to allocate weights to obtain the weights of each parameter. .

[0085] As an example, the on-orbit reliability is denoted as S, and its expression is: in, The parameters are weighted. The on-orbit reliability results are rounded to two decimal places, intuitively reflecting the overall operational reliability level of the electric thruster.

[0086] In summary, this technical solution enables multi-dimensional quantification and visualization of the on-orbit reliability of electric thrusters, accurately adapting to the on-orbit monitoring needs of propulsion systems. Single-parameter reliability can quickly pinpoint operational weaknesses: low thrust reliability indicates abnormal plasma generation and unstable propellant flow; low specific impulse reliability reflects deteriorating propulsion efficiency and abnormal ion acceleration channels; low power reliability indicates potential issues with power supply matching and energy transmission links, facilitating targeted troubleshooting and optimization by ground maintenance personnel. Individual unit reliability, as a comprehensive representation of the overall system's operational capability, indicates fewer parameter mutations and stronger operational stability with higher scores. It provides quantitative support for on-orbit mission scheduling, maintenance decisions, and parameter adjustments, reducing the risk of propulsion failures and ensuring the stable execution of spacecraft orbit and attitude control missions.

[0087] The comprehensive reliability is obtained by merging the overall compliance rate of performance parameters and the on-orbit operational reliability with the arithmetic mean. The reliability level is then determined based on the comprehensive reliability, thus obtaining the reliability evaluation result.

[0088] As an example, let the overall reliability be denoted as R, and its expression is: R = (Overall performance parameter compliance rate + On-orbit operational reliability) ÷ 2 The result should be rounded to two decimal places. Grading rules: 98≤R≤100, Excellent; 90≤R<98, good; 80≤R<90, moderate; 65≤R<80, poor; R < 65, range.

[0089] In summary, the basic indicators rely on long-term time-series statistics to reflect the continuous compliance of thrust, specific impulse, and power; the differential feature scheme can capture parameter mutations and provide early warning of sudden failures. The two are complementary in dimensions and can comprehensively characterize system reliability. After fusion, it can cover risks at multiple time scales, without the need for manual weighting, and can reflect both long-term operational stability and instantaneous fault early warning, meeting the reliability assessment requirements of electric thrusters.

[0090] 3. Stability As one possible implementation, see Figure 4 The stability evaluation results were obtained using the following method: (1) Stability of temperature control system + stability of attitude control system The thruster temperature, power processing unit temperature, thrust measurement unit temperature, and flow controller temperature from the thermal system parameters are obtained as a time series of temperature parameters, and the X-axis composite angular momentum, Y-axis composite angular momentum, and Z-axis composite angular momentum from the attitude control associated parameters are obtained as a time series of angular momentum parameters. Among them, the temperature stability of the electric thruster directly determines the thrust output accuracy, energy conversion efficiency, and on-orbit reliability. The thruster temperature characterizes the thermal state of the core propulsion components; the power processing unit (PPU) temperature characterizes the heat dissipation performance of the power processing unit; the temperature sensing unit (TSU) temperature characterizes the working environment of the sensing unit itself; and the flow controller temperature characterizes the thermal stability of the propellant delivery components.

[0091] Among them, the stability of the angular momentum of the attitude control system directly determines the attitude control accuracy, stability and on-orbit reliability. The fluctuation of the combined angular momentum of the X-axis, Y-axis and Z-axis, namely Hx, Hy and Hz, will affect the spacecraft's attitude pointing accuracy, the stability of the control torque output and the collaborative efficiency of the attitude control actuator.

[0092] As an example, the time series data acquisition of temperature, i.e., angular momentum parameters, uses a default time window of 1 hour per window, with each window containing multiple sampling points, which can be adjusted by the user as needed. The sampling mode defaults to matching the standard engineering configuration to meet the standard deviation calculation. Data cleaning removes extreme outliers to ensure calculation accuracy. For example, when a single temperature jump is greater than 8°C, the single angular momentum jump is greater than 0.05 N·m·s, and the previous sampling point is used to replace it. When a sampling point is missing, it is filled with the average of the data already collected in the window to ensure data integrity.

[0093] Calculate the standard deviations of temperature and angular momentum parameters over multiple time windows, and calculate the single-parameter stability scores of the temperature control system and the attitude control system based on the ratio of the standard deviation of each parameter to the preset stability threshold. Among them, the temperature stability threshold is preset according to the equipment model and working scenario; the angular momentum stability threshold is preset according to the spacecraft model, attitude control mode and working scenario.

[0094] As an example, the standard deviation of a single-window parameter: Where n > 30, n is taken as continuous data points according to the time series.

[0095] Mean of standard deviation for multi-window parameters: Calculate the arithmetic mean of the standard deviations of all windows for each parameter.

[0096] Single-parameter stability score: Score = 10 - 10 × r (r ≤ 1) Rating = 0 (r > 1) Where r is the ratio of the mean standard deviation of the parameter to the benchmark threshold.

[0097] The stability scores of the temperature control system are weighted and summed to obtain the stability score of the temperature control system. The weights are determined by the entropy weight method based on the dispersion of each temperature parameter. The stability scores of the single parameters of the attitude control system are aggregated by mean, and the stability score of the attitude control system is obtained. In summary, both attitude control and temperature stability solutions use standard deviation as the core indicator to achieve refined quantification and comprehensive evaluation of multi-dimensional parameter fluctuations in the corresponding systems. This overcomes the limitations of single-axis / single-parameter monitoring and aligns with the characteristics of three-axis coordination and dynamic balance of angular momentum in attitude control systems, as well as the multi-component coordination and close thermal state correlation in electric thruster temperature control systems. Both solutions capture short-term parameter fluctuations and reflect long-term stability levels through single / multi-window standard deviation and mean calculations. Combined with graded scoring rules and strong constraint mechanisms, they quickly determine whether parameter fluctuations exceed safe limits. Single / comprehensive scoring accurately locates weak points and reflects the overall stability level of the system. The temperature stability solution further introduces the entropy weight method to dynamically allocate weights, avoiding the subjectivity of manual weighting and conforming to the engineering realities of temperature control systems.

[0098] (3) On-orbit stability ①Support Vector Machine The time sequence of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as thermal system parameters, including thruster temperature, and propellant delivery parameters, including low-pressure values, is input into the trained support vector machine model to obtain state recognition labels, and the support vector machine score is calculated based on the state recognition labels. Among them, the Support Vector Machine (SVM) model is a classic supervised machine learning algorithm with strong small-sample learning ability and good generalization performance, which is suitable for binary classification evaluation of the operational stability of electric thrusters. This method finds the optimal hyperplane in the high-dimensional feature space to achieve "normal / abnormal" classification. The classification boundary is determined by the support vectors, which can resist the noise interference of telemetry data and is suitable for multi-parameter collaborative monitoring.

[0099] Among them, the bus current reflects the power input status, other electrical system current parameters reflect the power output and electrode working status of the electric propulsion system; the thruster temperature reflects the hot state of the core components; and the low-pressure value reflects the propellant delivery stability.

[0100] The status identification labels include: normal or abnormal; the classification results are highly robust and can effectively resist transient noise interference in the data.

[0101] As an example, the support vector machine score = max{0, (number of normal labels in the window / total number of labels) × 10 - (number of consecutive outliers × 0.5)} The number of labels is obtained through daily statistics from a sliding window; the support vector machine score quantifies the operational stability of the electric thruster on a 10-point scale. A higher score indicates smaller fluctuations and stronger stability in equipment operation, providing an intuitive quantitative basis for on-orbit operation and maintenance decisions.

[0102] ② Decision Tree The time sequence of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output current / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as thermal system parameters, including thruster temperature, and propellant delivery parameters, including low-pressure values, is input into the trained decision tree model to obtain state identification labels, and the decision tree score is calculated based on the state identification labels. Among them, the decision tree is a supervised machine learning algorithm with strong interpretability and high computational efficiency, which is suitable for rapid classification and evaluation of the operational stability of electric thrusters. The algorithm constructs a tree model by recursively dividing the feature space, and the nodes determine the classification based on a single feature threshold, outputting a "normal / abnormal" label, which is suitable for rapid state determination with multiple parameters.

[0103] Among them, various current / voltage parameters reflect the coordinated state of the electrical system; thruster temperature reflects thermal stability; low-pressure value reflects propellant delivery state, and multiple parameters provide sufficient basis for feature division of the decision tree.

[0104] The status identification labels include: normal or abnormal; the classification rules can be intuitively explained through a tree structure, making it easier for the ground team to understand the judgment logic.

[0105] As an example, the decision score is calculated as follows: 0, (number of normal labels within the window / total number of labels) × 10 - (number of consecutive anomalies × 0.5)} The number of labels is obtained through daily statistics from a sliding window; the decision score quantifies the operational stability of the electric thruster using a 10-point scale. A higher score indicates smaller fluctuations and stronger stability in equipment operation, and can be linked with support vector machine scoring to verify the stability assessment results.

[0106] ③K nearest neighbors The time series of parameters including the output current / voltage of the grid, the output current / voltage of the anode, the output current / voltage of the neutralizer heating, the acceleration output current, the excitation power supply current, and the bus current of the electrical system, the thruster temperature of the thermal system, and the low-pressure value of the propellant delivery parameters are input into the trained K-nearest neighbor model to obtain the matching similarity with the normal mode, and the K-nearest neighbor score is calculated based on the matching similarity. Among them, the K-Nearest Neighbors (KNN) model is a supervised machine learning algorithm that requires no training, is well-adapted to nonlinear data, and is suitable for the similarity matching evaluation of the state stability of electric thrusters. The algorithm calculates the similarity between the current sample and the normal mode sample, determines the state based on the majority vote of the K nearest neighbor samples, and outputs the matching similarity, which is suitable for multi-parameter mode matching requirements.

[0107] The matching similarity value ranges from 0 to 1. The closer the similarity is to 1, the closer the current state is to the normal pattern; the closer it is to 0, the more significantly the state deviates from the normal pattern.

[0108] As an example, the K-nearest neighbor score = state stability index × 10; Among them, the state stability index = similarity mean × (1 - coefficient of variation); the similarity mean, standard deviation and coefficient of variation are all obtained by sliding window statistics; the K-nearest neighbor score quantifies the state stability of the electric thruster on a 10-point scale. The higher the score, the higher the matching degree between the current state and the normal mode and the stronger the stability, providing a basis for early warning of state deterioration.

[0109] ④ Fluctuations in key performance parameters Calculate the coefficients of variation of thrust, specific impulse, and power, and calculate the fluctuation stability score of key performance parameters based on the coefficients of variation and the weights determined by the entropy weight method. Among them, the coefficient of variation (CV) is used to eliminate the difference in parameter dimensions, realize the unified fluctuation quantification across parameters, and complete the stability assessment without the need for preset thresholds, which is suitable for the needs of multi-parameter collaborative monitoring.

[0110] Among them, thrust, in mN, reflects the propulsion power output capability; specific impulse, in s, reflects the propulsion energy efficiency level; and power, in kW, reflects energy consumption and conversion efficiency.

[0111] As an example, the calculation process for the fluctuation stability score of key performance parameters is as follows: First, calculate the mean of each parameter within each time window, denoted as μ; the mean can reflect the overall level of the parameter during that period.

[0112] Furthermore, calculate the standard deviation of the parameter within each window, denoted as σ; the standard deviation quantifies the fluctuation range of the parameter from the mean, and the larger the standard deviation, the worse the operational stability.

[0113] Calculate the coefficient of variation (CV): CV = σ ÷ μ × 100%; Standard deviation normalization eliminates the influence of the mean on fluctuation assessment, enabling cross-parameter and cross-condition horizontal comparisons; simultaneously, if the thrust variation coefficient (CV) F A high specific impulse coefficient (CV) requires investigation into plasma generation and propellant flow abnormalities; I Large fluctuations may indicate issues with power supply output accuracy or ion acceleration channels; power coefficient of variation (CV) P A sudden increase requires checking the compatibility of the energy transmission link and the module.

[0114] The stability of the key performance parameter fluctuation is denoted as Y. in, The weights are used to indicate the trend of health decline. If Y continues to decline, an early warning can be triggered, providing a quantitative basis for adjusting power, starting backup thrusters, and planning maintenance during on-orbit operation and maintenance.

[0115] In summary, this technical solution can achieve standardized and threshold-free quantitative evaluation of multi-parameter fluctuations in electric thrusters, overcoming the limitations of single parameters and dimensional differences. By combining sliding window statistics, coefficient of variation, and entropy weight method, it can achieve stable parameter evaluation, eliminate dimensional barriers, give stability a clear physical meaning, and reduce the complexity of engineering applications.

[0116] (4) Fusion computing The stability scores of the temperature control system, attitude control system, support vector machine, decision tree, K-nearest neighbor, and key performance parameter fluctuation stability are fused by arithmetic mean to obtain a comprehensive stability score. The stability level is then determined based on the comprehensive stability score, thus obtaining the stability evaluation result.

[0117] As an example, the stability level classification rules are as follows: 9≤Q≤10, Excellent; 8 ≤ Q < 9, good; 6 ≤ Q < 8, moderate level; 3≤Q<6, qualified; 0 ≤ Q < 3, range.

[0118] Q represents the overall stability score.

[0119] In summary, this technical solution adopts a fusion approach to obtain stability evaluation. The four evaluation perspectives, namely basic indicators, neural networks, and data features, each have their own emphasis. Single-dimensional evaluation cannot cover all dimensions of stability characteristics. By integrating the advantages of various solutions, it can achieve single parameter fluctuation quantification, multi-parameter classification and identification, and cross-parameter normalized comparison. It does not require preset weights and comprehensively reflects the overall stability level of electric thruster temperature, attitude, and core performance fluctuations.

[0120] 4. Health status As one possible implementation, see Figure 5 The health status assessment results were obtained using the following methods: (1) Current health of electric thruster The electric thruster current health status is calculated based on the deviation of the current parameters from the rated current, including the grid output current, anode output current, anode holding output current, neutralizer holding output current, acceleration output current, excitation power supply current, and bus current. Excessive current deviation rate directly reflects abnormal plasma generation, increased electrode wear, power supply link mismatch or unstable excitation magnetic field. Long-term deviation will lead to decreased thrust output accuracy, electrode overheating and ablation, and reduced propellant ionization efficiency, ultimately causing electric thruster start-up and shutdown failures or interruption of on-orbit missions.

[0121] Among them, the current parameter acquisition sampling frequency is synchronized with the telemetry downlink frequency, for example, the sampling frequency is 1~10Hz to ensure real-time capture of current changes; lost telemetry data and current jump values ​​without cause are removed, such as single jump amplitude exceeding the rated current by 10% without external factors such as load adjustment; the proportion of valid sampling points within the data acquisition period is ≥98% to avoid evaluation bias caused by missing data.

[0122] As an example, the relative deviation rates of each current: Mean current deviation rate: Where M is the number of valid sampling points within the data acquisition period; the calculation logic for the average deviation rate of the seven types of currents is consistent.

[0123] Entropy weight method for calculating current weight. The entropy weight method is used to assign weights to the average deviation rate of each current, resulting in seven categories of key current weights. The weights reflect the degree of influence of the current deviation of each motor on the overall health. The motor with the greater deviation fluctuation has a higher weight ratio.

[0124] Current health status: in, As the weight, H I The health score ranges from 0 to 10. The closer the score is to 10, the better the health of the current system.

[0125] In summary, this technical solution captures the degree of deviation and long-term trend of current parameters by using real-time current deviation rate and average deviation rate, and combines this with a normalized health score to reflect the overall health level of the system. It can quickly locate abnormal current types and corresponding faults, and the health score can be linked with indicators such as temperature stability and thrust accuracy to construct a comprehensive health assessment system.

[0126] (2) Health status coupled with multiple temperature parameters The temperature of the thruster, power processing unit, thrust measurement unit, and flow controller in the thermal system are obtained. The correlation coefficients between each pair of temperature parameters are calculated, and a normalized health correlation matrix is ​​constructed based on the correlation coefficients. The coupling health of multiple temperature parameters is calculated based on the mean of the off-diagonal elements of the normalized health correlation matrix. Among them, the temperatures of the key components of the electric thruster exhibit strong coupling characteristics. The synergistic relationship between the temperature fields of the core components and the auxiliary systems is an important indicator of its health status. Stable coupling indicates that the thermal control, propulsion, and power processing systems work together efficiently, while weakened coupling indicates problems such as component wear, heat dissipation link failure, and propellant delivery thermal imbalance, which directly affect the thrust output accuracy and equipment life.

[0127] As an example, the parameters are collected by long-term continuous sampling with a total sampling length of ≥200 points to ensure the statistical significance of the correlation calculation; sampling is performed at equal time steps with data integrity ≥98%, and data from transient operating conditions such as equipment start-up and shutdown and power supply mode switching are excluded; each parameter sequence has no constant value, as constant value sequences cannot reflect coupling relationships and have no analytical significance. If a constant value exists, it should be marked and the coupling calculation of that parameter pair should be excluded.

[0128] As an example, first, construct the original correlation coefficient matrix R. in, Let r be the Pearson correlation coefficient between any two temperature parameters. ij ∈[−1,1],∣r ij The larger the value of |, the stronger the coupling between parameters. i,j=1-4 correspond to the four temperature parameters. Depend on Form a 4×4 original correlation coefficient matrix R.

[0129] Furthermore, the normalized health-related matrix R is obtained. health , R after normalization health (i,j)∈[0,1], the larger the value, the tighter the coupling between parameters, and the visualization shows the distribution of coupling strength between the four types of temperature parameters; Matrix normalization can eliminate the influence of positive and negative correlations, retaining only the coupling strength information.

[0130] Finally, the health status of the multi-temperature parameter coupling is calculated. Among them, H corr For health status coupled with multiple temperature parameters, H corr ∈[0,1], the closer the value is to 1, the more stable the thermo-electric coupling relationship is.

[0131] As an example, the health score is calculated by coupling multiple temperature parameters. ×10; It can normalize the scores with other algorithm schemes, and the score is quantified into 10 points to quantify the overall health level of the electric thruster temperature control system, and supports continuous time-series output to track trends.

[0132] In summary, this technical solution uses the coupling correlation of multiple temperature parameters as the core analysis index. By constructing a correlation matrix and performing thresholdless statistical operations, it achieves a quantitative evaluation of the high-order health status of electric thrusters, providing a precise basis for in-depth health assessment and fault location. For example, a decrease in the coupling degree between thruster temperature and power processing unit temperature indicates a thermal imbalance between the core propulsion components and the power unit, which may lead to problems such as blocked heat dissipation links and abnormal power matching. A weakening coupling degree between thrust measurement unit temperature and flow controller temperature can predict the thermal synergy failure of the sensing unit and propellant delivery system, thus avoiding the risk of thrust fluctuations in advance.

[0133] (3) Neural network health ①XGBoost Health Score The time series of cathode heating / ablation power supply output current, anode output current, anode output voltage, grid output current, grid output voltage, acceleration output current, neutralizer holding output current, neutralizer holding output voltage, excitation power supply current, DC / AC voltage, thruster temperature, bus current, and low-pressure value from multi-dimensional on-board telemetry data are input into the trained XGBoost model to obtain anomaly probability values, and the XGBoost health score is calculated based on the anomaly probability values. Among them, XGBoost is an ensemble learning algorithm with high training efficiency and strong generalization ability, which is suitable for accurate assessment of the health status of electric thrusters. The algorithm selects the optimal splitting feature through a greedy algorithm, avoids overfitting by combining regularization, fits the residual with a tree model, and outputs the abnormal probability of the sample, which is suitable for multi-parameter nonlinear correlation features.

[0134] Among them, various current / voltage parameters reflect the coordinated state of the electrical system; thruster temperature reflects thermal stability; and low-pressure value reflects the propellant delivery state.

[0135] The anomaly probability value ranges from 0 to 1. The closer the probability value is to 1, the higher the probability that the sample belongs to an abnormal state; the closer it is to 0, the closer the operating state is to normal.

[0136] As an example, XGBoost health score = (1 - anomaly probability value) × 10; The XGBoost health score quantifies the overall health level of the electric thruster on a 10-point scale. A higher score indicates a lower degree of degradation in core modules such as the electrical system and thermal control system, and a better health status.

[0137] ② Local outlier health score The time series of the screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, thruster temperature, power processing unit temperature, thrust measurement unit temperature, and low-pressure value from the multi-dimensional on-board telemetry data are input into the trained local outlier model to obtain the local outlier score, and the local outlier health score is calculated based on the local outlier score. Among them, the Local Outlier Factor (LOF) is an unsupervised outlier detection algorithm that does not require label data and can identify samples with abnormal local density, making it suitable for unsupervised assessment of the health status of electric thrusters. The algorithm quantifies the degree of outlier by calculating the ratio of the local density of a sample to that of its neighbors (LOF score). The higher the score, the more the sample deviates from the normal local density, making it suitable for label-free monitoring scenarios.

[0138] Among them, the Local Outlier Factor Score (LOF score) indicates that the higher the score, the more the current operating state of the electric thruster deviates from the local density of the normal sample; the lower the score, the closer the state is to normal.

[0139] As an example, the process of calculating the health score of the local outlier factor: The LOF scores of the training set are statistically analyzed, and the maximum value is taken as the normal threshold. This threshold serves as the benchmark for determining the normal state. Normalization process: If the LOF score is less than or equal to the normal threshold, the normalized value is 1 - (LOF score / normal threshold). If the LOF score is greater than the normal threshold, the normalized value is 0.5 - (LOF score - normal threshold) / (2 × normal threshold). Local outlier health score = normalized value × 10 The local outlier factor health score quantifies the health status of the electric thruster on a 10-point scale. The higher the score, the healthier the equipment is operating, providing a quantitative basis for operation and maintenance decisions in unsupervised scenarios.

[0140] ③ Health score of isolated forests The time series of the screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, thruster temperature, power processing unit temperature, thrust measurement unit temperature, and low-pressure value from the multi-dimensional on-board telemetry data are input into the trained isolated forest model to obtain the anomaly probability value, and the isolated forest health score is calculated based on the anomaly probability value. Among them, the Isolation Forest is an unsupervised outlier detection algorithm with high detection efficiency and good adaptability to high-dimensional data, making it suitable for efficient assessment of the health status of electric thrusters. The algorithm constructs multiple isolated trees by randomly dividing the feature space. The shorter the path of a sample in the tree, the easier it is to be isolated, i.e., the more abnormal it is. Finally, it is converted into anomaly probability values, which are suitable for the needs of efficient monitoring of multiple parameters.

[0141] The anomaly probability value ranges from 0 to 1. The closer the probability value is to 1, the higher the probability that the sample belongs to an abnormal state; the closer it is to 0, the closer the operating state is to normal.

[0142] As an example, the health score of an isolated forest = (1 - anomaly probability value) × 10; The score quantifies the overall health level of the electric thruster on a 10-point scale. A higher score indicates a lower degree of degradation in core modules such as the electrical system and thermal control system, and a better health status.

[0143] ④ Health score of one-dimensional convolutional neural network The time series of screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, and thruster temperature from multi-dimensional on-board telemetry data are input into a trained one-dimensional convolutional neural network model to obtain the health state probability, and the health score of the one-dimensional convolutional neural network is calculated based on the health state probability. Among them, the one-dimensional convolutional neural network (1D-CNN) is a deep learning algorithm that can extract local features of time series data, has high computational efficiency, and is suitable for deep time series assessment of the health status of electric thrusters. This algorithm captures local correlation features of time series data through one-dimensional convolutional layers, combines pooling layers to reduce dimensionality, and outputs health status probability values, which is suitable for monitoring needs of complex time series patterns.

[0144] Among them, various current / voltage parameters reflect the dynamic changes of the electrical system; thruster temperature reflects the thermal evolution.

[0145] The probability of health status ranges from 0 to 1. The closer the probability value is to 1, the better the health status of the electric thruster at the current moment; the closer it is to 0, the worse the health status.

[0146] As an example, the health score of a one-dimensional convolutional neural network = the probability of a healthy state × 10; The one-dimensional convolutional neural network health score quantifies the overall health level of the electric thruster on a 10-point scale. The higher the score, the lower the degree of degradation of core modules such as the electrical system and thermal control system, and the better the health status.

[0147] Extract the minimum value among the XGBoost health score, the local outlier health score, the isolated forest health score, and the one-dimensional convolutional neural network health score, and use it as the minimum health score of the neural network. The minimum health score of the neural network is denoted as H. min To capture the most significant risks of health deterioration and avoid underestimating potential risks with a single model.

[0148] (4) Fusion computing The arithmetic mean of the electric thruster current health status, multi-temperature parameter coupled health status, and neural network minimum health score is fused to obtain a comprehensive health score. The health level is then determined based on the comprehensive health score, thus obtaining the health status evaluation result.

[0149] As an example, the rules for classifying health levels are as follows: 9.0≤H≤10, healthy; 8.0≤H<9.0, sub-healthy; 5.0≤H<8.0, indicating poor health; 0≤H<0.5, fault (extremely poor); H represents the overall health score.

[0150] In summary, this technical solution integrates basic indicators, data characteristics, and neural network evaluation results, leveraging their respective advantages. It takes the minimum value of the neural network evaluation to strengthen the early warning of shortcomings, and then integrates it with the average value of the first two solutions to achieve a comprehensive health assessment of the electric thruster. To avoid the one-sidedness of a single assessment, there is no need to preset weights, and it can provide early warning of hidden degradation, taking into account both comprehensiveness and engineering practicality.

[0151] In a second aspect, embodiments of the present invention provide an on-orbit evaluation system for electric thrust, comprising: The data acquisition module acquires multi-dimensional on-board telemetry data of the electric thruster during its on-orbit operation. The multi-dimensional on-board telemetry data includes at least electrical system parameters, thermal system parameters, propellant delivery parameters, and attitude control related parameters. The performance evaluation module is used to evaluate the on-orbit performance of the electric thruster based on multi-dimensional on-board telemetry data and obtain the performance evaluation results. The reliability evaluation module is used to evaluate the reliability of the electric thruster based on multi-dimensional on-board telemetry data and obtain the reliability evaluation results. The stability evaluation module is used to evaluate the stability of the electric thruster based on multi-dimensional on-board telemetry data and obtain the stability evaluation results. The health status assessment module is used to assess the health status of the electric thruster based on multi-dimensional on-board telemetry data and obtain the health status assessment results. Among them, at least one of the performance evaluation module, reliability evaluation module, stability evaluation module and health status evaluation module is specifically used to: perform calculations using at least two of the basic index calculation unit, neural network prediction unit and data feature analysis unit, and fuse the calculation results to obtain the evaluation result of the evaluation dimension; The output module is used to output performance evaluation results, reliability evaluation results, stability evaluation results, and health status evaluation results.

[0152] As one possible implementation, the performance evaluation module includes: The basic performance calculation unit is used to calculate the time series of thrust, specific impulse and power based on the grid output voltage, grid output current, neutralizer contact voltage, neutralizer contact current, anode contact voltage, anode contact current, anode voltage, anode current, acceleration voltage, acceleration current and preset engineering parameters from multi-dimensional on-board telemetry data. It also calculates the overall on-orbit performance score of the electric thruster based on the comparison of the mean, range and standard deviation of thrust, specific impulse and power with the engineering calibration parameters. The first neural network prediction unit is used to input the time series of the grid output current, grid output voltage, anode output current, anode output voltage, neutralizer heating output current, neutralizer heating output voltage, neutralizer holding output current, neutralizer holding output voltage, acceleration output current, acceleration output voltage, excitation power supply current, DC / AC voltage, thruster temperature, bus current, and low-pressure value from the multi-dimensional on-board telemetry data into the trained long short-term memory network model, and calculate the electric thruster performance change trend score based on the abnormal probability value output by the model. The second neural network prediction unit is used to input the time series of screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, and thruster temperature from multi-dimensional on-board telemetry data into the trained deep autoencoder model, and calculate the electric thruster degradation trend score based on the reconstruction error output by the model. The performance fusion unit is used to fuse the on-orbit comprehensive performance score, performance change trend score, and degradation trend score of the electric thruster by arithmetic mean to obtain the comprehensive performance score.

[0153] As one possible implementation, the reliability evaluation module includes: The basic performance calculation unit is used to acquire thrust, specific impulse and power data during the ignition working period of the electric thruster, and calculate the overall compliance rate of performance parameters by comparing the data with the safety threshold range. The data feature analysis unit is used to calculate the first-order difference sequence of thrust, specific impulse and power during the working period of the electric thruster. By comparing the absolute value of the first-order difference sequence with the mutation threshold, the on-orbit reliability of the electric thruster is calculated. The reliability fusion unit is used to fuse the comprehensive compliance rate of performance parameters and the on-orbit operational reliability by arithmetic mean to obtain the comprehensive reliability.

[0154] Thirdly, embodiments of the present invention provide a terminal, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the on-orbit evaluation method for electric thrusters provided by the present invention.

[0155] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the description of the drawings, in carrying out the claimed invention. In this specification, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple components. A single processor or other unit can implement several of the functions listed in the specification. While certain measures are described in different embodiments, this does not mean that these measures cannot be combined to produce good results.

[0156] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely illustrative of the invention and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications fall within the scope of the invention and its equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A method for on-orbit evaluation of an electric thruster, characterized in that, The steps include the following: Acquire multi-dimensional on-board telemetry data of the electric thruster during its on-orbit operation. The multi-dimensional on-board telemetry data includes at least electrical system parameters, thermal system parameters, propellant delivery parameters, and attitude control related parameters. Based on multi-dimensional on-board telemetry data, at least two of the following methods are used to calculate the on-orbit performance, reliability, stability, and health status of the electric thruster: basic index calculation, neural network prediction, and data feature analysis. The calculation results are then fused to obtain on-orbit performance evaluation results, reliability evaluation results, stability evaluation results, and health status evaluation results.

2. The on-orbit evaluation method for electric thrusters according to claim 1, characterized in that, The on-orbit comprehensive performance score is obtained by averaging the on-orbit basic performance score, performance change trend score, and degradation trend score. The on-orbit performance level is determined based on the on-orbit comprehensive performance score, thus obtaining the on-orbit performance evaluation result. The on-orbit basic performance score is calculated using basic indicators, specifically including the following steps: Based on the electrical system parameters including the grid output voltage / current, neutralizer holding output voltage / current, anode holding output voltage / current, anode output voltage / current, acceleration output voltage / current, and preset engineering parameters, calculate the timing sequence of thrust, specific impulse, and power. Calculate the mean, range, and standard deviation of thrust, specific impulse, and power; Based on the comparison of the mean, range, and standard deviation of thrust, specific impulse, and power with the engineering calibration parameters, the mean compliance score, range compliance score, and fluctuation stability score of thrust, specific impulse, and power are calculated respectively. The mean compliance score, range compliance score, and fluctuation stability score of thrust, specific impulse, and power are aggregated to obtain the single-parameter on-orbit performance score of thrust, specific impulse, and power. The weights of thrust, specific impulse, and power are determined using the entropy weight method. The on-orbit performance scores of individual parameters are weighted and summed to obtain the overall on-orbit performance score of the electric thruster.

3. The on-orbit evaluation method for electric thrusters according to claim 2, characterized in that, Both the performance change trend score and the degradation trend score are obtained using neural network prediction, specifically including the following steps: The time series of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output current / voltage, acceleration output / voltage, excitation power supply current, DC / AC voltage, and bus current, as well as thermal system parameters, including thruster temperature, and thruster delivery parameters, including low-pressure values, are input into a trained long short-term memory network model to obtain the anomaly probability value for each time step. Calculate the performance change trend score of the electric thruster based on the anomaly probability value; The electrical system parameters, including the grid output current / voltage, anode output current / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as the thermal system parameters, including the thruster temperature, are input into the trained deep autoencoder model to obtain the reconstruction error. The degradation trend score of the electric thruster is calculated based on the reconstruction error.

4. The on-orbit evaluation method for electric thrusters according to claim 1, characterized in that, The reliability evaluation results were obtained using the following methods: Acquire thrust, specific impulse, and power data during the ignition and operation period of the electric thruster; The thrust, specific impulse, and power data are logically compared with their respective safety threshold ranges, and the number of normal sampling points for each parameter during the ignition operation period is counted. The compliance rates of thrust, specific impulse, and power are calculated based on the ratio of the number of normal sampling points to the total number of sampling points during the ignition operation period. The compliance rates of thrust, specific impulse, and power are averaged and aggregated to obtain the overall compliance rate of performance parameters; Calculate the first-order difference sequence of thrust, specific impulse, and power during the operating period of the electric thruster; The absolute value of the first-order difference sequence is compared with a preset mutation threshold, and the mutation number of each parameter is counted. Calculate the single-parameter reliability of each parameter based on the number of mutations; The weights of thrust, specific impulse, and power are determined by using the entropy weight method. The reliability of each single parameter is then weighted and summed to obtain the on-orbit reliability of the electric thruster. The comprehensive reliability is obtained by merging the overall compliance rate of performance parameters and the on-orbit operational reliability with the arithmetic mean. The reliability level is then determined based on the comprehensive reliability, thus obtaining the reliability evaluation result.

5. The on-orbit evaluation method for electric thrusters according to claim 1, characterized in that, The stability evaluation results were obtained using the following method: The thruster temperature, power processing unit temperature, thrust measurement unit temperature, and flow controller temperature from the thermal system parameters are obtained as a time series of temperature parameters, and the X-axis composite angular momentum, Y-axis composite angular momentum, and Z-axis composite angular momentum from the attitude control associated parameters are obtained as a time series of angular momentum parameters. Calculate the standard deviations of temperature and angular momentum parameters over multiple time windows, and calculate the single-parameter stability scores of the temperature control system and the attitude control system based on the ratio of the standard deviation of each parameter to the preset stability threshold. The stability scores of the temperature control system are weighted and summed to obtain the stability score of the temperature control system. The weights are determined by the entropy weight method based on the dispersion of each temperature parameter. The stability scores of the single parameters of the attitude control system are aggregated by mean, and the stability score of the attitude control system is obtained. The time sequence of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as thermal system parameters, including thruster temperature, and propellant delivery parameters, including low-pressure values, is input into the trained support vector machine model to obtain state identification labels, and the support vector machine score is calculated based on the state identification labels. The time sequence of electrical system parameters, including grid output current / voltage, anode output current / voltage, neutralizer heating output current / voltage, neutralizer holding output current / voltage, acceleration output current / voltage, excitation power supply current, and bus current, as well as thermal system parameters, including thruster temperature, and propellant delivery parameters, including low-pressure values, is input into the trained decision tree model to obtain state identification labels, and the decision tree score is calculated based on the state identification labels. The time series of parameters including the output current / voltage of the grid, the output current / voltage of the anode, the output current / voltage of the neutralizer heating, the acceleration output current, the excitation power supply current, and the bus current of the electrical system, the thruster temperature of the thermal system, and the low-pressure value of the propellant delivery parameters are input into the trained K-nearest neighbor model to obtain the matching similarity with the normal mode, and the K-nearest neighbor score is calculated based on the matching similarity. Calculate the coefficients of variation of thrust, specific impulse, and power, and calculate the fluctuation stability score of key performance parameters based on the coefficients of variation and the weights determined by the entropy weight method. The stability scores of the temperature control system, attitude control system, support vector machine, decision tree, K-nearest neighbor, and key performance parameter fluctuation stability are fused by arithmetic mean to obtain a comprehensive stability score. The stability level is then determined based on the comprehensive stability score, thus obtaining the stability evaluation result.

6. The on-orbit evaluation method for electric thrusters according to claim 1, characterized in that, The health status assessment results were obtained using the following methods: The electric thruster current health status is calculated based on the deviation of the current parameters from the rated current, including the grid output current, anode output current, anode holding output current, neutralizer holding output current, acceleration output current, excitation power supply current, and bus current. The temperature of the thruster, power processing unit, thrust measurement unit, and flow controller in the thermal system are obtained. The correlation coefficients between each pair of temperature parameters are calculated, and a normalized health correlation matrix is ​​constructed based on the correlation coefficients. The coupling health of multiple temperature parameters is calculated based on the mean of the off-diagonal elements of the normalized health correlation matrix. The time series of cathode heating / ablation power supply output current, anode output current, anode output voltage, grid output current, grid output voltage, acceleration output current, neutralizer holding output current, neutralizer holding output voltage, excitation power supply current, DC / AC voltage, thruster temperature, bus current, and low-pressure value from multi-dimensional on-board telemetry data are input into the trained XGBoost model to obtain anomaly probability values, and the XGBoost health score is calculated based on the anomaly probability values. The time series of the screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, thruster temperature, power processing unit temperature, thrust measurement unit temperature, and low-pressure value from the multi-dimensional on-board telemetry data are input into the trained local outlier model to obtain the local outlier score, and the local outlier health score is calculated based on the local outlier score. The time series of multi-dimensional on-board telemetry data, including grid output current, grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, thruster temperature, power processing unit temperature, thrust measurement unit temperature, and low-pressure value, are input into the trained isolated forest model to obtain anomaly probability values, and the isolated forest health score is calculated based on the anomaly probability values. The time series of screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, and thruster temperature from multi-dimensional on-board telemetry data are input into a trained one-dimensional convolutional neural network model to obtain the health state probability, and the health score of the one-dimensional convolutional neural network is calculated based on the health state probability. Extract the minimum value among the XGBoost health score, the local outlier health score, the isolated forest health score, and the one-dimensional convolutional neural network health score, and use it as the minimum health score of the neural network. The arithmetic mean of the electric thruster current health status, multi-temperature parameter coupled health status, and neural network minimum health score is fused to obtain a comprehensive health score. The health level is then determined based on the comprehensive health score, thus obtaining the health status evaluation result.

7. An on-orbit evaluation system for electric thrust, characterized in that, include: The data acquisition module acquires multi-dimensional on-board telemetry data of the electric thruster during its on-orbit operation. The multi-dimensional on-board telemetry data includes at least electrical system parameters, thermal system parameters, propellant delivery parameters, and attitude control related parameters. The performance evaluation module is used to evaluate the on-orbit performance of the electric thruster based on multi-dimensional on-board telemetry data and obtain the performance evaluation results. The reliability evaluation module is used to evaluate the reliability of the electric thruster based on multi-dimensional on-board telemetry data and obtain the reliability evaluation results. The stability evaluation module is used to evaluate the stability of the electric thruster based on multi-dimensional on-board telemetry data and obtain the stability evaluation results. The health status assessment module is used to assess the health status of the electric thruster based on multi-dimensional on-board telemetry data and obtain the health status assessment results. Among them, at least one of the performance evaluation module, reliability evaluation module, stability evaluation module and health status evaluation module is specifically used to: perform calculations using at least two of the basic index calculation unit, neural network prediction unit and data feature analysis unit, and fuse the calculation results to obtain the evaluation result of the evaluation dimension; The output module is used to output the performance evaluation results, reliability evaluation results, stability evaluation results, and health status evaluation results.

8. The electric thrust on-orbit evaluation system according to claim 7, characterized in that, The performance evaluation module includes: The basic performance calculation unit is used to calculate the time series of thrust, specific impulse and power based on the grid output voltage, grid output current, neutralizer contact voltage, neutralizer contact current, anode contact voltage, anode contact current, anode voltage, anode current, acceleration voltage, acceleration current and preset engineering parameters from multi-dimensional on-board telemetry data. It also calculates the overall on-orbit performance score of the electric thruster based on the comparison of the mean, range and standard deviation of thrust, specific impulse and power with the engineering calibration parameters. The first neural network prediction unit is used to input the time series of the grid output current, grid output voltage, anode output current, anode output voltage, neutralizer heating output current, neutralizer heating output voltage, neutralizer holding output current, neutralizer holding output voltage, acceleration output current, acceleration output voltage, excitation power supply current, DC / AC voltage, thruster temperature, bus current, and low-pressure value from the multi-dimensional on-board telemetry data into the trained long short-term memory network model, and calculate the electric thruster performance change trend score based on the abnormal probability value output by the model. The second neural network prediction unit is used to input the time series of screen grid output current, screen grid output voltage, anode output current, anode output voltage, acceleration output current, acceleration output voltage, excitation power supply current, bus current, and thruster temperature from multi-dimensional on-board telemetry data into the trained deep autoencoder model, and calculate the electric thruster degradation trend score based on the reconstruction error output by the model. The performance fusion unit is used to fuse the on-orbit comprehensive performance score, performance change trend score, and degradation trend score of the electric thruster by arithmetic mean to obtain the comprehensive performance score.

9. The electric thrust on-orbit evaluation system according to claim 7, characterized in that, The reliability evaluation module includes: The basic index calculation unit is used to acquire thrust, specific impulse and power data during the ignition working period of the electric thruster, and calculate the comprehensive compliance rate of performance parameters by comparing the data with the safety threshold range. The data feature analysis unit is used to calculate the first-order difference sequence of thrust, specific impulse and power during the working period of the electric thruster. By comparing the absolute value of the first-order difference sequence with the mutation threshold, the on-orbit reliability of the electric thruster is calculated. The reliability fusion unit is used to fuse the comprehensive compliance rate of performance parameters and the on-orbit operational reliability by arithmetic mean to obtain the comprehensive reliability.

10. A terminal, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the on-orbit evaluation method for an electric thruster as described in any one of claims 1 to 6.

Images