A nano-ampere current resolution-based ion pump adaptive control method

Abstract

The present application relates to the technical field of ion pump, and particularly relates to an ion pump adaptive control method based on nano-ampere current resolution; through sliding window dynamic threshold and mutation waveform classification, transient current rising spikes caused by normal starting and gas emission of the ion pump and abnormal discharge spikes can be accurately distinguished, unnecessary protection actions caused by misjudgment of fixed threshold in the starting stage are effectively avoided, through the use of second-order difference acceleration discrimination condition, non-physical pulses such as electromagnetic interference are effectively filtered out to interfere with the rising spike detection, the false positive rate is significantly reduced, and through the automatic adjustment of threshold coefficients by closed-loop statistical feedback, the detection sensitivity is adaptively matched to different noise working conditions, the long-term operation stability is high, and the three-stage soft protection timing designed for oscillation abnormalities is used to reduce the voltage with controllable slope, maintain low voltage to eliminate ionization, and slowly increase the voltage to recover, so that the voltage impact and secondary sparking risk caused by hard shutdown are avoided, and the discharge continuity is maintained while the micro-discharge is inhibited.

Description

Technical Field

[0001] This invention relates to the field of ion pump technology, and in particular to an adaptive control method for ion pumps based on nanoampere-level current resolution. Background Technology

[0002] Ion pumps (sputter ion pumps) utilize the Penning discharge principle to generate ion current. The amplitude and trend of this ion current directly reflect the pressure state of the vacuum system and the operating status of the ion pump itself. During operation, the controller typically monitors the ion current to assess the system vacuum level in real time and executes corresponding control and protection logic accordingly. Existing ion pump controllers generally use the comparison between the real-time collected current and a fixed threshold as an indicator for subsequent control and protection. For example, a fixed overcurrent protection threshold is set; when the ion current exceeds this threshold, the controller determines it as an abnormal discharge and automatically cuts off the high-voltage output or triggers a voltage reduction protection mechanism.

[0003] However, at the moment of ion pump startup, the desorption of residual gas inside the pump and gas adsorbed on the cathode surface causes instantaneous gas release, resulting in a sudden jump in ion current from the nanoampere background noise level to several microamperes or even tens of microamperes, forming a transient current spike. This is a normal phenomenon during the normal startup process of an ion pump. However, for ion pump controllers using fixed threshold detection, it is difficult to accurately distinguish between the transient current spike caused by normal startup gas release and the abnormal current spike caused by a sudden drop in vacuum or internal arcing. The former is an inherent phenomenon of the normal startup process, while the latter is an abnormal operating condition that requires intervention. Existing fixed threshold methods make indiscriminate judgments on both using the same standard, which may lead to the transient current spike during normal startup being misjudged as abnormal discharge, causing the controller to perform unnecessary protection operations. This problem is particularly prominent in applications that require frequent start-stop of ion pumps, such as ion pumps in particle accelerators and semiconductor coating equipment. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an adaptive control method for ion pumps based on nanoampere-level current resolution. This method solves the technical problem that existing ion pump controllers using fixed threshold detection cannot accurately distinguish between transient current spikes caused by normal start-up venting and abnormal current spikes caused by sudden drops in vacuum or internal arcing, which leads to misjudgments by the ion pump controller.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: an adaptive control method for an ion pump based on nanoampere-level current resolution, used for state monitoring and control of an ion pump controller. This method specifically includes the following steps: S1. Real-time acquisition of ion current signals from the ion pump to construct a current sampling sequence; S2. Set the sliding window with the sampling point of the previous moment as the corresponding sliding window at the current moment, and calculate the mean and standard deviation of the current value of each sampling point in the sliding window. S3. Calculate the high dynamic threshold and low dynamic threshold of the current sampling point based on the mean and standard deviation of the current value corresponding to the current sliding window. S4. Generate the status label of the ion pump at the current sampling point based on the current of the ion pump at the previous sampling point, the high dynamic threshold, and the low dynamic threshold. If the current of the ion pump at the current sampling point is greater than the high dynamic threshold, and the change in the current of the ion pump at the current sampling point relative to the previous sampling point is greater than a preset multiple of the standard deviation, the generated label is a rising spike abrupt change. If the current of the ion pump at the current sampling point is less than the low dynamic threshold, and the change in the current of the ion pump at the current sampling point relative to the previous sampling point is less than the negative of the preset multiple of the standard deviation, the generated label is a falling step change. If the number of alternating rising spikes and falling step changes in multiple consecutive sampling points reaches a preset number of alternations, the generated label is oscillation anomaly.

[0006] S5. Generate a corresponding control command to the high-voltage power supply module based on the current status label of the ion pump to adjust its output voltage.

[0007] Preferably, in step S2, the length of the sliding window is determined based on the current operating voltage of the ion pump, and when the current operating voltage of the ion pump is within a preset rated operating voltage range, the length of the sliding window is negatively correlated with the operating voltage of the ion pump within a preset window length range.

[0008] Preferably, in step S4, the discrimination rule for the sudden change of the rising peak also includes: the difference between the current change at the current sampling point of the ion pump and the current change at the previous sampling point is a positive value.

[0009] Preferably, in step S4, the rising spike mutation includes at least a first rising spike subtype and a second rising spike subtype; When a sudden rise peak is detected within the preset time window after the ion pump starts, it is determined to be the first rise peak subtype. In step S5, when the current sampling point is detected to be the first rise peak subtype, the controller keeps the output voltage of the high voltage power supply module unchanged. When a sudden rise peak is detected outside the preset time window after the ion pump starts, it is determined to be the second rise peak subtype. In step S5, when the current sampling point is detected to be the second rise peak subtype, the controller triggers the high voltage sag protection to reduce the output voltage to the preset protection voltage value.

[0010] Preferably, in step S5, when the status label is oscillation abnormal, the control command is to control the high-voltage power supply module to execute the corresponding control measures in sequence according to the timing of the descent phase, the holding phase and the recovery phase. The control measure for the falling phase is to reduce the output voltage from the current voltage to the preset protection voltage at a preset falling slope, and the duration of the falling phase is less than or equal to a preset falling time threshold. The control measures during the holding phase are to maintain the output voltage at a preset protection voltage for a preset holding time. The control measures during the recovery phase are to restore the output voltage to the output voltage at the start of the falling phase with a preset rising slope, and the rise time is within a preset rise time range.

[0011] Preferably, in step S3, the high dynamic threshold of the current sampling point is determined by calculating the product of the standard deviation of the current value corresponding to the sliding window at the current moment and a positive high dynamic threshold coefficient, and by summing the product with the mean of the current value corresponding to the sliding window. The low dynamic threshold of the current sampling point is determined by calculating the product of the standard deviation of the current value corresponding to the sliding window at the current moment and a positive low dynamic threshold coefficient, and by the difference between the mean of the current value corresponding to the sliding window and the product.

[0012] Preferably, the high dynamic threshold coefficient and the low dynamic threshold coefficient are calculated based on the cumulative number of false positive events and false negative events within a preset statistical period, and the specific steps are as follows: A false positive event is defined as follows: when the generated status label is a sudden rise in the peak or an abnormal oscillation, after the high-voltage power supply module executes the corresponding control command, the ion current fails to recover to the current safe range centered on the sliding window mean within the preset evaluation time window. A false negative event is defined as follows: when the generated status label is normal, the ion current shows a sudden increase in peak value within a preset evaluation time window and the duration reaches a preset duration threshold. The cumulative number of false positive events detected within a preset statistical period is counted. and the number of false negative events And calculate the proportion of each of the two in the total number of tests. The proportion of false positives is used to obtain the false positive rate. and false negative rate ; Based on false positive rate and the preset false positive rate threshold and false negative rate and the preset false negative rate threshold The high dynamic threshold coefficient and low dynamic threshold coefficient are adjusted according to the following rules for calculation in the next statistical period: like > and < If the false alarm rate is too high, the high dynamic threshold coefficient m is increased. like ≤ and > If the number of false negatives is too high, the high dynamic threshold coefficient will be reduced. like > and > If the noise and sudden change interference are both detected, then the high dynamic threshold coefficient and the low dynamic threshold coefficient should be increased. Otherwise, the high dynamic threshold coefficient and the low dynamic threshold coefficient remain unchanged.

[0013] The present invention also provides a system for an adaptive control method for an ion pump, comprising a processor and a memory, wherein the memory is used to store a computer program, which, when executed by the processor, implements the adaptive control method for the ion pump.

[0014] By employing the above technical solution, the present invention provides an adaptive control method for ion pumps based on nanoampere-level current resolution, which has at least the following beneficial effects: 1. This invention uses a sliding window dynamic threshold and abrupt waveform classification to accurately distinguish between transient current rise spikes and abnormal discharge spikes caused by normal start-up venting of the ion pump. This effectively avoids unnecessary protection actions triggered by misjudgment of fixed thresholds during the start-up phase, ensuring smooth ignition of the ion pump and shortening the start-up time.

[0015] 2. This invention utilizes second-order differential acceleration discrimination conditions to effectively filter out interference from non-physical pulses such as electromagnetic interference on the detection of rising spikes, significantly reducing the false positive rate; at the same time, it automatically adjusts the threshold coefficient through closed-loop statistical feedback, so that the detection sensitivity adaptively matches different noise conditions, resulting in high long-term operational stability.

[0016] 3. This invention designs a three-stage soft protection timing sequence of "decline-hold-recovery" for abnormal oscillations. It uses a controllable slope to reduce voltage, maintain low voltage to deionize, and slowly increase voltage to recover. This avoids voltage surges and secondary arcing risks caused by hard shutdown. It maintains discharge continuity while suppressing micro-discharges, protects the ion pump electrodes, and improves system reliability. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of the adaptive control method for ion pumps based on nanoampere-level current resolution according to the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This will allow for a full understanding of how the present application uses technical means to solve technical problems and achieve technical effects, and to facilitate its implementation.

[0019] Example 1: To address the technical problem that existing ion pump controllers using fixed threshold detection struggle to accurately distinguish between transient current spikes caused by normal start-up venting and abnormal current spikes caused by sudden vacuum drops or internal arcing, leading to frequent misjudgments, this example provides an adaptive control method for ion pumps based on nanoampere-level current resolution. This method is used for state monitoring and control of the ion pump controller. Taking a 400 L / s sputtering ion pump as an example, its operating voltage range is 3 kV to 10 kV. The ion current detection system uses a 24-bit Σ-Δ ADC with a sampling rate set to 2 kHz to ensure resolution of nanoampere-level current signals and real-time capture of transient changes. The method specifically includes the following steps: S1. Real-time acquisition of the ion current signal from the ion pump to construct a current sampling sequence. In this embodiment, the ion current formed between the anode and cathode during Penning discharge is extremely weak, especially under ultra-high vacuum conditions (such as...). Below Pa, the ion current can be as low as nanoamperes. In order to effectively distinguish the current signal reflecting the vacuum level change and discharge state from the noisy substrate, this embodiment uses a high-precision analog-to-digital converter (ADC) with a fixed sampling period. The ion current signal is continuously acquired, with a sampling rate of 2kHz. Each sampling moment yields a discrete current value. For example, the current value at the k-th sampling moment, or the k-th sampling point, can be denoted as... Where k=1,2,3,… are the sampling point numbers, arranging all sampling points in chronological order forms the current sampling sequence for subsequent calculations. .

[0020] S2. The original current sampling sequence obtained in step S1 may contain background noise, power frequency interference, and possible actual current changes. If a fixed threshold comparison is directly performed on each sampling point, it is very easy to cause misjudgment due to noise spikes. Therefore, in this embodiment, a sliding window with a length of N and an endpoint at the sampling point of the previous moment is set as the sliding window corresponding to the current moment. The length of the sliding window can be set to 100, which corresponds to a time width of 50ms. If the current moment is the kth sampling point, the starting point of the sliding window is the kNth sampling point, and the ending point is the (k-1)th sampling point. That is, the sliding window contains N consecutive historical sampling points before the current sampling point, but does not include the current sampling point itself. All sampling points in the sliding window constitute a subset of background noise. Then, the arithmetic mean of the current values ​​at each sampling point in the sliding window is calculated, which is used as an estimate of the DC level of the current background noise. The standard deviation of the current values ​​at each sampling point in the sliding window is also calculated as an estimate of the fluctuation amplitude of the current background noise.

[0021] S3. Based on the mean and standard deviation of the current value corresponding to the current sliding window, calculate the high dynamic threshold and low dynamic threshold of the current sampling point. In this embodiment, the high dynamic threshold of the current sampling point is determined by multiplying the standard deviation of the current value corresponding to the current sliding window at the current moment with a positive high dynamic threshold coefficient m, and by summing the product with the mean of the current value corresponding to the sliding window. The calculation formula can be expressed as:

[0022] In the above formula, This represents the high dynamic threshold corresponding to the k-th sampling point. and This represents the mean and standard deviation of the current values ​​at each sampling point within the sliding window corresponding to the k-th sampling point.

[0023] The low dynamic threshold for the current sampling point is determined by multiplying the standard deviation of the current value corresponding to the sliding window at the current moment with a positive low dynamic threshold coefficient n, and by the difference between the mean of the current value corresponding to the sliding window and the product. Therefore, the formula for calculating the low dynamic threshold corresponding to the k-th sampling point can be expressed as:

[0024] Where m and n are used to control the sensitivity of the high dynamic threshold and the low dynamic threshold, respectively.

[0025] S4, the high dynamic threshold, and the low dynamic threshold provide the upper and lower limits of the normal range of the current. However, comparing these two thresholds alone is insufficient to distinguish between normal startup transient spikes and abnormal discharge faults, because the current value in both cases may exceed the high dynamic threshold. To differentiate the waveform characteristics of these two cases, this embodiment generates the current status label of the ion pump at the current sampling point, the current of the ion pump at the previous sampling point, the high dynamic threshold, and the low dynamic threshold as the judgment result. Before the judgment, a parameter can be calculated. For the k-th sampling point, the current value of the current k-th sampling point is first calculated. The current value relative to the previous, i.e., the (k-1)th sampling point Change The calculation formula can be expressed as:

[0026] Then, status labels are generated according to the following discrimination rules. : If the current of the ion pump at the current sampling point is greater than the high dynamic threshold, i.e. And when the change in current of the ion pump at the current sampling point relative to the previous sampling point is greater than a preset multiple of the standard deviation, i.e. The preset multiplier can be set to 3, and the generated label is a rising spike mutation, which can be denoted as... This judgment rule requires that not only does the current exceed the high dynamic threshold, but the rate of rise is also significantly greater than the current noise fluctuation level, thereby eliminating misjudgments caused by instantaneous noise pulses. This embodiment can further add the following condition: the difference between the current change at the current sampling point of the ion pump and the current change at the previous sampling point is positive. That is, when these three conditions are met simultaneously, the generated label is a sudden rise peak. The essence of this difference is the direction of the current change, which is the second-order difference between the current at the current sampling point and the current at the previous sampling point. This is used to distinguish the monotonically accelerating current spike generated by the ion pump's start-up venting from the instantaneous pulses produced by electromagnetic interference or random noise. Specifically, at the moment of ion pump start-up, due to the desorption of gas adsorbed on the cathode surface, the ion current exhibits a physically regular rising waveform: this rising process is not a step, but rather has an accelerating rising phase (i.e., the change in current itself is continuously increasing), which is mathematically reflected as a first-order difference. As time increases, the corresponding second-order difference becomes positive. Conversely, the rising edge of instantaneous pulses generated by electromagnetic interference or ADC sampling anomalies often exhibits a single-point jump, i.e. If the value of a sample suddenly increases from a small value at the previous sampling point and then quickly falls back, its second-order difference often oscillates between positive and negative values ​​and will not remain positive.

[0027] In the ion pump startup experiment of this embodiment, the sampling rate was 2kHz (sampling interval 0.5ms), the sliding window N=100, and the current background noise was... High dynamic threshold The original current values ​​at three consecutive sampling points are: , , Then the calculation yields:

[0028]

[0029]

[0030] Therefore, the following three conditions are met:

[0031]

[0032]

[0033] Therefore, it was determined to be an upward spike mutation, and a state label was generated. .

[0034] If the current of the ion pump at the current sampling point is less than the low dynamic threshold, i.e. And when the change in current of the ion pump at the current sampling point relative to the previous sampling point is less than the negative of the preset multiple of the standard deviation, i.e. The generated label is a descending step transition, which can be denoted as: ; If the number of alternating rising spikes and falling step changes within multiple consecutive sampling points reaches a preset number of alternations, the generated label is oscillation anomaly. In this embodiment, if the number of alternating rising spikes and falling step changes within 5 consecutive sampling points reaches 3 times, the generated label is oscillation anomaly, which can be denoted as... This mode typically corresponds to micro-discharge or arcing phenomena, characterized by violent positive and negative oscillations of current over a short period of time.

[0035] If any of the above conditions are not met, it is considered normal and can be recorded as follows: .

[0036] S5. When the current exceeds the high dynamic threshold, it may correspond to different reasons such as normal startup or abnormal ignition. Therefore, it is necessary to execute corresponding control commands based on the tag. Thus, in this embodiment, the ion pump's current status tag generates a corresponding control command to the high-voltage power supply module to adjust its output voltage. In this embodiment, the controller (using an STM32F407 microcontroller) can... Different values ​​output different control commands: like When a sudden rise in voltage spike is detected, this embodiment further classifies it into at least two types: a first rise in voltage spike subtype and a second rise in voltage spike subtype. When a sudden rise in voltage spike is detected within a preset time window (3 seconds in this embodiment) after the ion pump starts, it is determined to be the first rise in voltage spike subtype. At this time, it is determined to be a transient current rise in voltage spike caused by normal start-up venting. When the current sampling point is detected to be the first rise in voltage spike subtype in step S5, the controller keeps the output voltage of the high-voltage power supply module unchanged. When a sudden rise in voltage spike is detected outside the preset time window after the ion pump starts, it is determined to be the second rise in voltage spike subtype. At this time, it can be identified as an abnormal current rise in voltage spike caused by factors such as a sudden drop in vacuum or internal arcing. When the current sampling point is detected to be the second rise in voltage spike subtype in step S5, the controller triggers high-voltage sag protection, reducing the output voltage to a preset protection voltage value. In this embodiment, the protection voltage value is set to 70% of the current operating voltage, that is, the voltage drop is 30%, the voltage drop rate is generally controlled at 100V / ms, and the drop time does not exceed 10ms. Because the ion pump will generate transient current spikes during normal startup, this is a normal phenomenon. The outgassing phenomenon after the ion pump starts up is mainly concentrated in the first 1 to 5 seconds after startup. The duration of outgassing varies slightly depending on the pump model, size, electrode material, and initial pressure of the vacuum system. Experimental verification shows that for most sputtering ion pumps (pumping speed 100L / s to 1000L / s), 3 seconds is a reasonable value that can cover the main outgassing stage without affecting the detection of abnormalities. Those skilled in the art can adjust this parameter within the range of 1 to 5 seconds according to the specific pump model.

[0037] like When a step change occurs, it usually indicates a rapid increase in vacuum (such as when a valve is closed). The controller maintains the current voltage and keeps the system stable using only conventional PID control.

[0038] like When an oscillation occurs abnormally, the current waveform exhibits high-frequency, large-amplitude positive and negative oscillations, indicating that the discharge has lost stability. Without timely intervention, this can lead to electrode surface damage, vacuum degradation, or even pump failure. Traditional controllers typically immediately cut off the high-voltage output when faced with oscillation anomalies, waiting several seconds before attempting to restart. This abrupt approach has the following drawbacks: ① The voltage surge caused by the hard shutdown may trigger induced overvoltage, exacerbating arcing; ② After completely cutting off the high voltage, the discharge is completely extinguished, requiring the pump to undergo a restart process, resulting in a long recovery time; ③ Frequent hard shutdowns accelerate the aging of high-voltage switching devices. Therefore, the control command in this embodiment is to control the high-voltage power supply module to execute corresponding three-stage control measures sequentially according to the falling phase, holding phase, and recovery phase. That is, instead of completely cutting off the high voltage, the voltage is temporarily reduced to a level sufficient to maintain a weak discharge but insufficient to sustain arcing, held for a period to allow the discharge products to dissipate, and then slowly restored to the original voltage. The entire process is smooth and controllable, suppressing arcing while maintaining the continuity of the discharge. The control measures for the falling phase are as follows: the output voltage is reduced from the current voltage to a preset protection voltage, such as 70% of the maximum rated voltage, at a preset falling slope. The duration of the falling phase is less than or equal to a preset falling time threshold, which is generally set to 10ms. The control measures for the holding phase are as follows: the output voltage is maintained at a preset protection voltage for a preset holding time, which is generally set to 50-200ms. The control measures for the recovery phase are as follows: the output voltage is restored to the output voltage at the beginning of the falling phase at a preset rising slope, and the rising time is within a preset rising time range, such as 30-60ms.

[0039] like Under normal conditions, the controller maintains the current output voltage and allows for regular voltage closed-loop regulation.

[0040] This embodiment also provides a system for an adaptive control method for an ion pump, including a processor and a memory. The memory is used to store a computer program, which is executed by the processor to implement the adaptive control method for the ion pump.

[0041] Example 2 differs from Example 1 in that, in step S2, the length of the sliding window is determined based on the current operating voltage of the ion pump. When the current operating voltage of the ion pump is within a preset rated operating voltage range, the length of the sliding window is negatively correlated with the operating voltage of the ion pump within the preset window length range. Specifically, in this example, the operating voltage of the ion pump typically varies between 3kV and 10kV, and the noise characteristics of the ion current signal differ significantly at different voltage levels. When the operating voltage is low, such as 3kV to 5kV, the Penning discharge intensity is weak, the ion current amplitude is small, and the signal-to-noise ratio is low, with background noise, including power supply ripple and electromagnetic interference, being relatively prominent. In this case, to accurately estimate the statistical characteristics of the background noise, a larger sliding window length is needed to smooth out high-frequency noise and obtain a stable and reliable mean and standard deviation. Conversely, at higher operating voltages, such as 8kV to 10kV, discharge is enhanced, the ion current amplitude increases, and the signal-to-noise ratio improves. Furthermore, the controller needs to respond quickly to current surges (especially spikes caused by arcing). If the window length is too large, the background model will have an excessively long historical memory of the surges, delaying the response of the mean and standard deviation, thus reducing the sensitivity to subsequent surges. Therefore, under high-voltage conditions, the sliding window length should be reduced to improve the algorithm's response speed. Therefore, this embodiment uses a piecewise linear function negatively correlated with the operating voltage to determine the sliding window length N, as follows: Let the rated operating voltage range of the ion pump be... In this embodiment, , The maximum value of the preset window length range The minimum value is 200. The value is 50. When the actual operating voltage of the ion pump is within the rated operating voltage range, the window length N is calculated according to the following formula:

[0042] in, This indicates rounding down to the nearest integer.

[0043] When the actual operating voltage of the ion pump is lower than or higher If, for example, the voltage has not yet reached the rated value during the startup phase, then the following methods are used: and As a conservative value.

[0044] In Example 3, the high dynamic threshold coefficient *m* and the low dynamic threshold coefficient *n* directly determine the width of the dynamic threshold, thus affecting the detection sensitivity and the balance between false alarms and false negatives. Fixed values ​​for *m* and *n* cannot adapt to all operating conditions. If the coefficients are set too small, the detection sensitivity is too high, easily misreporting noise or normal fluctuations as sudden changes (i.e., false positives as defined below). If the coefficients are set too large, the detection sensitivity is too low, potentially missing true sudden changes (i.e., false negatives as defined below). Therefore, Example 3 was further designed to optimize the values ​​of *m* and *n* online by periodically evaluating the false positive and false negative rates, ensuring that the detection performance remains consistently near the optimal balance point.

[0045] Example 3 differs from Example 2 in that the high dynamic threshold coefficient and low dynamic threshold coefficient are calculated based on the cumulative number of false positive events and false negative events within a preset statistical period. The specific steps are as follows: First, a false positive event is defined as follows: When the generated status label indicates a sudden rise in voltage spike or abnormal oscillation, and the high-voltage power supply module executes the corresponding control command, the ion current fails to recover to the current safety range centered on the sliding window mean within the preset evaluation time window. This indicates that the controller detected a sudden rise in voltage spike or abnormal oscillation and executed corresponding control actions, such as voltage holding or voltage dip protection. However, if this situation is actually a normal fluctuation or non-destructive disturbance, and the control action is unnecessary, it is not considered a false positive.

[0046] A false negative event is defined as follows: when the generated status label is normal, the ion current exhibits a sudden upward spike within a preset evaluation time window, and the duration reaches a preset duration threshold. This indicates a missed detection; it was not detected during real-time monitoring and was considered normal, but a sudden upward spike was found in the current graph afterward, and the duration reached the preset duration threshold, such as 1.5ms. Therefore, it is a false negative event.

[0047] The cumulative number of false positive events detected within a preset statistical period is counted. and the number of false negative events And calculate the proportion of each of the two in the total number of tests. The proportion of false positives is used to obtain the false positive rate. and false negative rate Its calculation formula can be expressed as:

[0048]

[0049] Based on false positive rate and the preset false positive rate threshold and false negative rate and the preset false negative rate threshold The high dynamic threshold coefficient and low dynamic threshold coefficient are adjusted according to the following rules for calculation in the next statistical period: like > and < If the false alarms are too many, the high dynamic threshold coefficient m is increased, for example by increasing it by 5% or fixing the step size to 0.2. like ≤ and > If the false negative rate is too high, the high dynamic threshold coefficient will be reduced, for example by 5% or by a fixed step size of 0.2. like > and > If the noise and sudden change are both interferences, then the high dynamic threshold coefficient and the low dynamic threshold coefficient should be increased, such as by increasing them simultaneously by 5% or by fixing the step size to 0.2. Otherwise, the high dynamic threshold coefficient and the low dynamic threshold coefficient remain unchanged.

[0050] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0051] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Since the above embodiments are substantially similar to the method embodiments, their descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0052] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An adaptive control method for an ion pump based on nanoampere-level current resolution, used for state monitoring and control of an ion pump controller, characterized in that, The method specifically includes the following steps: S1. Real-time acquisition of ion current signals from the ion pump to construct a current sampling sequence; S2. Set the sliding window with the sampling point of the previous moment as the corresponding sliding window at the current moment, and calculate the mean and standard deviation of the current value of each sampling point in the sliding window. S3. Calculate the high dynamic threshold and low dynamic threshold of the current sampling point based on the mean and standard deviation of the current value corresponding to the current sliding window. S4. Generate the status label of the ion pump at the current sampling point based on the current of the ion pump at the previous sampling point, the high dynamic threshold, and the low dynamic threshold. If the current of the ion pump at the current sampling point is greater than the high dynamic threshold, and the change in the current of the ion pump at the current sampling point relative to the previous sampling point is greater than a preset multiple of the standard deviation, the generated label is a rising spike abrupt change. If the current of the ion pump at the current sampling point is less than the low dynamic threshold, and the change in the current of the ion pump at the current sampling point relative to the previous sampling point is less than the negative of the preset multiple of the standard deviation, the generated label is a falling step change. If the number of alternating rising spikes and falling step changes in multiple consecutive sampling points reaches the preset number of alternations, the generated label is oscillation anomaly; S5. Generate a corresponding control command to the high-voltage power supply module based on the current status label of the ion pump to adjust its output voltage.

2. The ion pump adaptive control method according to claim 1, characterized in that, In step S2, the length of the sliding window is determined based on the current operating voltage of the ion pump, and when the current operating voltage of the ion pump is within a preset rated operating voltage range, the length of the sliding window is negatively correlated with the operating voltage of the ion pump within a preset window length range.

3. The ion pump adaptive control method according to claim 1, characterized in that, In step S4, the discrimination rule for the sudden change of the rising peak also includes: the difference between the current change at the current sampling point of the ion pump and the current change at the previous sampling point is a positive value.

4. The ion pump adaptive control method according to claim 1, characterized in that, In step S4, the rising spike mutation includes at least a first rising spike subtype and a second rising spike subtype; When a sudden rise peak is detected within the preset time window after the ion pump starts, it is determined to be the first rise peak subtype. In step S5, when the current sampling point is detected to be the first rise peak subtype, the controller keeps the output voltage of the high voltage power supply module unchanged. When a sudden rise peak is detected outside the preset time window after the ion pump starts, it is determined to be the second rise peak subtype. In step S5, when the current sampling point is detected to be the second rise peak subtype, the controller triggers the high voltage sag protection to reduce the output voltage to the preset protection voltage value.

5. The ion pump adaptive control method according to claim 1, characterized in that, In step S5, when the status label is oscillation abnormal, the control command is to control the high-voltage power supply module to execute the corresponding control measures in sequence according to the timing of the descent phase, the holding phase and the recovery phase; The control measure for the falling phase is to reduce the output voltage from the current voltage to the preset protection voltage at a preset falling slope, and the duration of the falling phase is less than or equal to a preset falling time threshold. The control measures during the holding phase are to maintain the output voltage at a preset protection voltage for a preset holding time. The control measures during the recovery phase are to restore the output voltage to the output voltage at the start of the falling phase with a preset rising slope, and the rise time is within a preset rise time range.

6. The ion pump adaptive control method according to claim 1, characterized in that, In step S3, the high dynamic threshold of the current sampling point is determined by calculating the product of the standard deviation of the current value corresponding to the sliding window at the current moment and a positive high dynamic threshold coefficient, and by summing the product with the mean of the current value corresponding to the sliding window. The low dynamic threshold of the current sampling point is determined by calculating the product of the standard deviation of the current value corresponding to the sliding window at the current moment and a positive low dynamic threshold coefficient, and by the difference between the mean of the current value corresponding to the sliding window and the product.

7. The ion pump adaptive control method according to claim 6, characterized in that, The high dynamic threshold coefficient and low dynamic threshold coefficient are calculated based on the cumulative number of false positive events and false negative events within a preset statistical period. The specific steps are as follows: A false positive event is defined as follows: when the generated status label is a sudden rise in the peak or an abnormal oscillation, after the high-voltage power supply module executes the corresponding control command, the ion current fails to recover to the current safe range centered on the sliding window mean within the preset evaluation time window. A false negative event is defined as follows: when the generated status label is normal, the ion current shows a sudden increase in peak value within a preset evaluation time window and the duration reaches a preset duration threshold. The cumulative number of false positive events detected within a preset statistical period is counted. and the number of false negative events And calculate the proportion of each of the two in the total number of tests. The proportion of false positives is used to obtain the false positive rate. and false negative rate ; Based on false positive rate and the preset false positive rate threshold and false negative rate and the preset false negative rate threshold The high dynamic threshold coefficient and low dynamic threshold coefficient are adjusted according to the following rules for calculation in the next statistical period: like > and < If the false alarm rate is too high, the high dynamic threshold coefficient m is increased. like ≤ and > If the number of false negatives is too high, the high dynamic threshold coefficient will be reduced. like > and > If the noise and sudden change interference are both detected, then the high dynamic threshold coefficient and the low dynamic threshold coefficient should be increased. Otherwise, the high dynamic threshold coefficient and the low dynamic threshold coefficient remain unchanged.

8. A system for implementing the adaptive control method for an ion pump according to any one of claims 1-7, characterized in that, It includes a processor and a memory, the memory being used to store a computer program that, when executed by the processor, implements the ion pump adaptive control method as described in any one of claims 1-7.

Images