A method for measuring a liquid level based on a millimeter wave radar

Abstract

The application belongs to the technical field of liquid level measurement, and specifically discloses a liquid level measurement method based on millimeter wave radar, which comprises the following steps: performing radar detection for multiple times in continuous multiple measurement time windows to obtain original measurement value sequences of the windows; extracting the range and the change amount per unit time of the sequences to determine the liquid surface fluctuation level corresponding to the current measurement time window; performing fluctuation trend analysis on the liquid surface fluctuation levels of the continuous multiple measurement time windows; determining the adjustment direction of the window length according to the current liquid surface fluctuation level and the change trend; generating a sampling window length adjustment instruction of the next measurement time window and adjusting the sampling window length; and performing radar detection in the adjusted measurement time window to output a liquid level measurement result. Through dynamic adjustment of the sampling window length, the application overcomes the contradiction between response speed and stability caused by a single fixed window length, and realizes adaptive optimization of measurement performance.

Description

Technical Field

[0001] This invention belongs to the field of liquid level measurement technology, and more specifically, relates to a liquid level measurement method based on millimeter-wave radar. Background Technology

[0002] Millimeter-wave radar is widely used in the field of industrial liquid level measurement due to its advantages such as narrow beam, high resolution, and strong anti-interference capability.

[0003] In the prior art, various improvement schemes have emerged to improve measurement accuracy. For example, Chinese invention patent application No. 2021114032277 discloses a liquid level measurement method and radar liquid level gauge, which performs real-time compensation by measuring the tilt angle of the radar to eliminate the influence of installation errors. Another example is a marine guided wave radar liquid level measurement device and measurement method disclosed in Chinese invention patent application No. 2023103785436, which adopts an external measuring cylinder and a reference collar structure, combined with density difference compensation and tilt angle compensation, to further improve measurement accuracy and achieve high-precision liquid level measurement.

[0004] However, the aforementioned technologies all operate under a fixed measurement mode, meaning that the radar's sampling window length and other parameters remain constant. In practical industrial applications, liquid surface conditions exhibit dynamic changes; the liquid surface may suddenly transition from a calm state to a violent fluctuation, or gradually stabilize from a fluctuating state. A fixed measurement mode cannot simultaneously achieve optimal performance under both conditions, resulting in either a sluggish response or decreased measurement stability when conditions change.

[0005] Furthermore, compensation methods can only correct systematic errors in measurement results and cannot resolve the contradiction between dynamic response and stability caused by the mismatch between measurement mode and operating conditions. Therefore, it is difficult to balance response speed and measurement accuracy when operating conditions change. Summary of the Invention

[0006] In view of this, in order to solve the above problems, a liquid level measurement method based on millimeter-wave radar is proposed.

[0007] The objective of this invention can be achieved through the following technical solution: This invention provides a liquid level measurement method based on millimeter-wave radar, which includes: performing multiple radar detections within a series of consecutive measurement time windows to obtain an original measurement value sequence composed of multiple detection results within each measurement time window.

[0008] Extract the range and change per unit time from the original measurement sequence to determine the liquid level fluctuation level corresponding to the current measurement time window.

[0009] Fluctuation trend analysis was performed on the liquid surface fluctuation levels over multiple consecutive measurement time windows to obtain the changing trend of the liquid surface fluctuation state. The changing trend can be one of the following: tending to stabilize, fluctuation intensification, or remaining unchanged.

[0010] The direction for adjusting the window length is determined based on the level of liquid level fluctuation within the current measurement time window and the observed trend.

[0011] Adjust the direction based on the window length, generate the sampling window length adjustment command for the next measurement time window, and execute the measurement time window adjustment.

[0012] Radar detection is performed within the adjusted measurement time window to output the corresponding liquid level measurement results.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention extracts the range and unit time change from the original measurement value sequence, and constructs the fluctuation feature set to quantify the current fluctuation relative index to determine the liquid surface fluctuation level by combining the historical data of multiple consecutive measurement time windows. It can accurately distinguish the three working conditions of stable, moderate fluctuation and violent fluctuation without relying on the preset fluctuation model, and provides an accurate working condition identification basis for the switching of measurement mode.

[0014] (2) By detecting the change points of the fluctuation level sequence, the present invention identifies the trend of fluctuation intensification or stabilization represented by the continuous rise or fall of the fluctuation level. This allows the direction of adjustment of the sampling window length to be predicted in advance based on the evolution trend of the working condition rather than just responding to the current state. This solves the problem of delayed response of the fixed measurement mode when the working condition changes suddenly, and ensures the adaptability of the sampling window setting.

[0015] (3) This invention generates a sampling window length adjustment command based on the joint judgment result of fluctuation level and change trend, and introduces the current measurement dispersion and echo quality factor to correct the adjustment coefficient during the adjustment process. This shortens the measurement time window length under severe fluctuation conditions to improve response speed and extends it under stable conditions to enhance measurement stability. Thus, it can accommodate the different requirements of the two conditions for measurement parameters within the same measurement cycle, overcoming the contradiction that a single fixed window length cannot take into account both response speed and stability.

[0016] (4) In the process of adjusting the window length, the present invention constrains the step size of the length change between adjacent measurement time windows, thereby avoiding drastic changes in window length caused by misjudgment of fluctuation level or abnormal echo quality, ensuring a smooth transition in the measurement parameter adjustment process, and further improving the control reliability and measurement results continuity under complex working conditions. Attached Figure Description

[0017] Figure 1This is a schematic diagram of the overall implementation process of the method of the present invention;

[0018] Figure 2 This is a schematic diagram of the liquid level fluctuation level process of the present invention;

[0019] Figure 3 This is a schematic diagram of the sampling window length adjustment instruction generation process of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] In practical industrial applications, liquid level conditions often exhibit dynamic changes, such as a sudden shift from a calm state to violent fluctuations, or a gradual stabilization from a fluctuating state. Using a fixed sampling window length cannot respond to diverse liquid level fluctuations. For instance, while a long sampling window can improve stability through averaging multiple measurements, it exhibits a sluggish response during liquid level fluctuations and cannot quickly track changes in liquid level. Conversely, a short sampling window, although responding quickly, has a low signal-to-noise ratio when the liquid level is stable, and the measured values ​​are easily affected by random noise.

[0022] To address the aforementioned issues, this embodiment proposes a liquid level measurement method based on millimeter-wave radar. This method first performs multiple radar detections within each measurement time window to acquire a sequence of raw measurement values, and extracts the range and change per unit time from this sequence as fluctuation characteristics. Secondly, the current liquid level fluctuation level is determined by comparing the fluctuation characteristics of the current window with those of historical windows. Thirdly, change point detection and trend analysis are performed on the fluctuation levels of multiple consecutive historical windows to determine whether the trend of liquid level fluctuation is towards stability, drastic fluctuation, or no change. Then, considering the current fluctuation level and the trend, the direction for adjusting the sampling window length of the next measurement time window is determined. Next, an adjustment coefficient is calculated based on the dispersion of the measured values ​​within the current window to generate the sampling window length for the next measurement time window. Finally, radar detection is performed within the adjusted window, and the final liquid level measurement result is output. Through these steps, the sampling window is automatically extended to enhance measurement stability when the liquid level is stable, and automatically shortened to improve tracking speed when the liquid level fluctuates drastically, thereby achieving adaptive optimization of measurement performance.

[0023] Please see details. Figure 1 As shown, Figure 1The present invention provides a liquid level measurement method based on millimeter-wave radar, which specifically includes the following steps: S1, performing multiple radar detections within multiple consecutive measurement time windows to obtain the original measurement value sequence composed of multiple detection results within each measurement time window.

[0024] It should be understood that the multiple consecutive measurement time windows are divided according to a fixed duration. The duration of each measurement time window can be set to, for example, 1 second, 2 seconds or 5 seconds, and can be preset according to the expected fluctuation frequency of the liquid surface being measured.

[0025] Specifically, within each measurement time window, the millimeter-wave radar continuously transmits detection signals at a preset sampling frequency and receives corresponding echo signals, thereby obtaining a sequence of original measurement values ​​consisting of multiple liquid level measurements arranged in the detection time sequence. This sequence is used to characterize the dynamic change process of the liquid level within the measurement time window.

[0026] S2. Extract the range and change per unit time from the original measurement value sequence to determine the liquid level fluctuation level corresponding to the current measurement time window.

[0027] Specifically, please refer to Figure 2 As shown, the process for determining the liquid level fluctuation level is as follows: S21, extract the range and the change per unit time from the original measurement value sequence corresponding to the current measurement time window. The range reflects the maximum fluctuation amplitude of the liquid level measurement value within the window, and the change per unit time reflects the average rate of change of the liquid level within the window. The extracted range and change per unit time constitute the current fluctuation feature pair.

[0028] S22. Obtain N consecutive historical measurement time windows preceding the current measurement time window, where N is a preset number of historical windows. The value of N is positively correlated with the accuracy requirement of the fluctuation level determination; that is, the higher the accuracy requirement, the larger the value of N should be. For example, N can be set to 50, 100, or 200. For each historical measurement time window, extract its range and the change per unit time in the same way as in step S21 to form a corresponding historical fluctuation feature pair, thereby obtaining a set containing N historical fluctuation feature pairs.

[0029] S23. Compare the range and change per unit time of the current fluctuation feature pair with the corresponding parameters of each historical fluctuation feature pair in the historical fluctuation feature pair set. Count the number of historical fluctuation feature pairs in the set that simultaneously satisfy both the condition that the range is less than the current range and the change per unit time is less than the current change, and record this statistical result as the first count value.

[0030] S24. Calculate the ratio of the first count value to the total number of feature pairs (i.e., N) in the historical fluctuation feature pair set, and use the obtained ratio as the current fluctuation relative index.

[0031] It should be noted that the relative volatility index ranges from 0 to 1, reflecting the rank of the current window's volatility relative to historical windows under current historical conditions. Specifically, the smaller the current relative volatility index, the more stable the volatility characteristics (range and change per unit time) of the current measurement window are in historical data, indicating that the current liquid level is in a relatively volatile state. Conversely, the larger the current relative volatility index, the more stable the volatility characteristics of the current window are in historical data, indicating that the current liquid level is in a relatively stable state.

[0032] S25. Based on the current fluctuation relative index value range, the liquid surface fluctuation level is divided into three levels: stable level, moderate fluctuation level, and severe fluctuation level.

[0033] The specific method for determining the threshold is as follows: During the calibration phase before or at the initial stage of the radar level gauge is officially put into use, the range and change per unit time of M consecutive measurement time windows are collected to form a set of historical fluctuation characteristics for calibration, where the value of M is not less than 100, for example, it can be set to 200, 300 or 500.

[0034] For each historical fluctuation feature pair in the calibration set, it is used as the current fluctuation feature pair. The calibration set of historical fluctuation feature pairs itself is used as the historical fluctuation feature pair set. The relative fluctuation index is calculated according to the method described in steps S22 to S23 above, resulting in M ​​calibrated relative fluctuation index values, forming a calibration index sequence. The calibration index sequence is then sorted in ascending order to obtain the sorted calibration index sequence.

[0035] The value at the 20th percentile in the sorted calibration index sequence is used as the first threshold. The value at the 80th percentile is used as the second threshold. Therefore, the bottom 20% of the calibrated index sequence corresponds to a relatively volatile state, while the top 80% corresponds to a relatively stable volatile state.

[0036] Based on the above thresholds, the rules for determining the level of liquid surface fluctuation are as follows: when the current fluctuation relative index is less than the first dividing threshold, the level of liquid surface fluctuation is determined to be a severe fluctuation level.

[0037] When the current fluctuation relative index is greater than the second threshold, the liquid surface fluctuation level is determined to be stable.

[0038] When the current relative volatility index is between the first and second thresholds, the liquid surface volatility level is determined to be moderate.

[0039] It should be noted that the 20% and 80% quantile values ​​mentioned above are only one specific implementation method in this embodiment, and can be adaptively adjusted according to measurement requirements in practical applications. In situations where high measurement stability is required, the values ​​can be appropriately increased. The value of can be chosen, for example, the 85th percentile as the second threshold, making the criteria for determining a stable level more stringent, thus obtaining a longer sampling window and more stable measurement results under stable operating conditions. In situations where high response speed is required, the value of can be appropriately reduced, for example, the 15th percentile as the first threshold, so that the sampling window can be shortened more quickly when fluctuations occur, thereby improving the response speed.

[0040] This step extracts the range and unit time change from the original measurement value sequence, and constructs a fluctuation feature set by combining historical data from multiple consecutive measurement time windows to quantify the current fluctuation relative index to determine the liquid surface fluctuation level. It can accurately distinguish between three working conditions: stable, moderate fluctuation and severe fluctuation, without relying on a preset fluctuation model, and provides an accurate working condition identification basis for switching measurement modes.

[0041] S3. Perform fluctuation trend analysis on the liquid surface fluctuation levels of multiple consecutive measurement time windows to obtain the changing trend of the liquid surface fluctuation state. The changing trend is one of the following: tending to be stable, fluctuation intensifies, or remains unchanged.

[0042] Specifically, the steps for performing fluctuation trend analysis are as follows: S31, obtain the liquid level fluctuation levels for the K consecutive measurement time windows preceding the current measurement time window, where K is the preset trend analysis window length, for example, it can be set to 10, 20 or 30. Arrange these fluctuation levels sequentially according to time order to form a fluctuation level sequence.

[0043] S32. Based on the fluctuation level sequence, perform change point detection, determine the direction of fluctuation level change after the most recent change point according to the change point detection results, and determine the fluctuation trend based on the change direction: when the direction of fluctuation level change after the most recent change point is continuously rising, the trend of the liquid surface fluctuation state is judged to be increased fluctuation.

[0044] When the fluctuation level after the most recent point of change is continuously decreasing, the trend of the liquid surface fluctuation is judged to be stabilizing.

[0045] When no change point is detected or the direction of change of the fluctuation level after the most recent change point remains unchanged, the trend of change of the liquid surface fluctuation state is judged to remain unchanged.

[0046] The specific steps for detecting change points in step S32 are as follows: First, the fluctuation level is converted into a fluctuation score, resulting in a fluctuation score sequence. In this embodiment, the stable level, moderate fluctuation level, and severe fluctuation level correspond to values ​​of 1, 2, and 3, respectively. The higher the fluctuation level, the larger the fluctuation score. This conversion method is only one example; other methods can be used in practical applications, and this invention does not limit the method.

[0047] Then, according to the preset sliding window width The sequence of fluctuation scores is slid sequentially from the beginning to the end. For example, it can be set to 5 or 8. At each sliding position... (in, The range of values ​​is from the first... Location to The first mean of the fluctuation score within the sliding window and the second mean of the fluctuation score outside the sliding window are respectively measured at each position. To the The first average is the average of the fluctuation scores, and the second average is the average of the remaining fluctuation scores in the fluctuation score sequence excluding the data within the window.

[0048] Next, iterate through all sliding positions, calculate the absolute difference between the first mean and the second mean, and mark the sliding position where the absolute difference reaches the maximum value as a candidate change point.

[0049] The absolute difference at the candidate change point is compared with a preset difference threshold. The preset difference threshold can be set, for example, to the 25th percentile of the theoretical maximum fluctuation score difference. For instance, in this embodiment, the theoretical maximum fluctuation score difference is 2, and the preset difference threshold can be set to 0.5. When the absolute difference is greater than the preset difference threshold, the candidate change point is output as a detected change point. Otherwise, it is considered that no valid change point has been detected.

[0050] In practical applications, if the fluctuation score sequence lacks candidate change points where the absolute difference between the means inside and outside the sliding window exceeds a preset threshold, no valid change points can be detected across the entire length. In this case, when the sequence length K exceeds a preset length threshold, change points must be forcibly marked to ensure the continuity of trend analysis. This involves marking the start or end point of the fluctuation score sequence as a change point.

[0051] Understandably, to ensure sufficient statistical samples both inside and outside the window when forcibly marking change points, the preset length threshold should be no less than twice the width of the sliding window, for example, it can be set to 10. Whether to mark the start or end point as the change point can be determined based on the direction of change at the end of the fluctuation score sequence. If the end of the sequence shows an upward trend, then the end point is marked as the change point. If the end of the sequence shows a downward trend, then the start point is marked as the change point. Otherwise, the end point is marked as the change point.

[0052] It should be noted that an upward trend at the end of the sequence means the slope of the linear fit of the last three fluctuation scores is positive. A downward trend at the end of the sequence means the slope of the linear fit of the last three fluctuation scores is negative. It should also be noted that the above method is only an example; implementers may use other methods, such as comparing the difference between the last two points or calculating the change in the mean at the end, to determine the trend.

[0053] Furthermore, in step S32, when determining the direction of fluctuation level change of the fluctuation level sequence after the most recent change point based on the change point detection results, if the starting point is marked as a change point, then the starting point is used as the starting point of the sequence analysis, and the direction of change of all subsequent fluctuation scores is analyzed; if the ending point is marked as a change point, then the subsequence before the ending point is used as the analysis object, and when the length of the subsequence is less than the preset length threshold, the change trend is determined to remain unchanged.

[0054] This step detects changes in the fluctuation level sequence to identify trends of intensified or stabilized fluctuations, characterized by continuous increases or decreases. This allows the direction of sampling window length adjustment to be predicted in advance based on the evolution of the operating conditions, rather than simply responding to the current state. This solves the problem of delayed response in fixed measurement modes when operating conditions change abruptly, and ensures the adaptability of sampling window settings.

[0055] S4. Determine the direction for adjusting the window length based on the level of liquid level fluctuation in the current measurement time window and the determined trend of change.

[0056] Specifically, the process for determining the direction of adjustment is as follows: S41, obtain the level of liquid surface fluctuation in the current measurement time window.

[0057] S42. When the liquid level fluctuation tends to stabilize, if the current measurement time window is at the stable level or the moderate fluctuation level, the adjustment direction is to extend the window; if the current measurement time window is at the severe fluctuation level, the adjustment direction is to maintain the window.

[0058] S43. When the liquid level fluctuation becomes more intense, adjust the direction to shorten the window.

[0059] S44. When the liquid level fluctuation remains unchanged, the adjustment direction is to maintain the window.

[0060] Understandably, when the trend is towards stabilization, it indicates that the level of liquid surface fluctuation is decreasing. In this case, for stable and moderate fluctuation levels, extending the sampling window helps to obtain more raw measurement values ​​and improve measurement stability. For severe fluctuation levels, considering that the current fluctuation is still relatively high, it is not advisable to suddenly extend the window; therefore, the window length should remain unchanged. When the trend is towards increased fluctuation, it indicates that the level of liquid surface fluctuation is increasing. In this case, regardless of the current fluctuation level, the sampling window should be shortened to improve the response speed to rapidly changing conditions. When the trend is towards no change, it indicates that the liquid surface fluctuation is relatively stable, and the current sampling window length should remain unchanged.

[0061] S5. Adjust the direction according to the window length, generate the sampling window length adjustment command for the next measurement time window, and execute the measurement time window adjustment.

[0062] Specifically, please refer to Figure 3 As shown, the sampling window length adjustment instruction for generating the next measurement time window specifically includes: S51, obtaining the standard deviation of multiple original measurement values ​​within the current measurement time window as the current measurement dispersion.

[0063] S52. Determine the adjustment coefficient based on the current measurement dispersion and the adjustment direction of the window length, and use the product of the current measurement time window length and the adjustment coefficient as the target window length.

[0064] S53. Based on the adjustment direction and target window length, obtain the sampling window length of the next measurement time window: When the adjustment direction is to extend the window, the smaller of the target window length and the preset maximum window length is used as the sampling window length of the next measurement time window.

[0065] When the adjustment direction is to shorten the window, the larger of the target window length and the preset minimum window length is used as the sampling window length for the next measurement time window.

[0066] When the orientation is adjusted to hold the window, the target window length is used as the sampling window length for the next measurement time window.

[0067] S54. Based on the sampling window length of the next measurement time window, generate a sampling window length adjustment command for the next measurement time window.

[0068] The following supplementary explanations are needed regarding the specific execution of steps S51 and S53: Before step S52, an echo quality verification step is performed to avoid incorrect window length adjustment due to calculation errors in the adjustment coefficient when the echo quality is poor. The specific verification process is as follows: Obtain the peak-to-sidelobe ratio of the radar echo for each frame within the current measurement time window and calculate its average value as the current echo quality factor. The peak-to-sidelobe ratio (PSNR) is the ratio of the peak value of the main lobe to the peak value of the maximum sidelobe in a radar echo signal. A higher PNR indicates better echo quality.

[0069] The current echo quality factor is compared with the preset echo quality warning threshold, which is usually set to 10dB. If the current echo quality factor is less than the preset echo quality warning threshold, it indicates that the current echo quality is poor. The adjustment direction is then corrected to maintain the window, and the adjustment coefficient is set to 1, that is, the sampling window length of the next measurement time window is consistent with the current window length.

[0070] If the current echo quality factor is greater than or equal to the preset echo quality warning threshold, then maintain the original adjustment direction and the echo quality verification passes.

[0071] Once the verification passes or the direction is corrected and adjusted, continue executing step S52.

[0072] It should be noted that the correction priority of the echo quality verification is higher than the adjustment direction judgment based on the liquid surface fluctuation level and change trend in step S4.

[0073] Furthermore, the specific calculation process of the adjustment coefficient in step S52 is as follows: S521, calculate the ratio of the current measurement dispersion to the preset benchmark dispersion to obtain the current dispersion fluctuation factor. .

[0074] Understandably, the preset baseline dispersion is updated in the following way: the measurement time window in which the liquid level fluctuation is stable and the trend of change remains unchanged is used as the target window, which represents that the liquid level is in a stable state.

[0075] For each target window, the standard deviation of multiple original measurements within that window is calculated as a sample of stationary standard deviations. This yields a sample of each stationary standard deviation. The mean of each sample of stationary standard deviations is used as the updated preset benchmark dispersion, so that the preset benchmark dispersion tracks the stationary dispersion of the liquid surface under different periods and environments, thereby improving the accuracy of the adjustment coefficient calculation.

[0076] It should be noted that during the initial measurement phase of the millimeter-wave radar, the standard deviation of the original measurement values ​​within a preset number of initial measurement time windows is used as the initial benchmark dispersion, and the preset number is not less than 10.

[0077] S522, Calculate the adjustment coefficient , , To set a reference echo quality factor, it is typically set to 15 dB, which represents the desired echo quality level. , and Each sets conditions. This indicates that the direction of adjustment is to extend the window. This indicates that the adjustment direction is to shorten the window. This indicates that adjusting the orientation will keep the window open. This represents the function that takes the maximum value. This represents the function that takes the minimum value.

[0078] in, To extend window branches, This indicates the echo quality correction term. This indicates a decrease in echo quality. When echo quality is poor, the signal-to-noise ratio (SNR) needs to be improved by extending the window to increase time-domain accumulation. Therefore, the larger this term is, the greater the window extension. Conversely, when echo quality is good... At this point, the extension range should be reduced.

[0079] At the same time, if This indicates that the current dispersion is higher than the baseline, and the liquid level is fluctuating wildly. Therefore, the window needs to be extended to smooth out the fluctuations. The larger the value, the greater the extension; conversely, if... This indicates that the current dispersion is lower than or equal to the baseline. At this point, the liquid level tends to be stable, and there is no need to extend it. This factor is less than 1, which means that the extension range is reduced.

[0080] This represents the product of the echo quality correction term and the discrete fluctuation correction term. The product of these two terms comprehensively reflects the two factors that require an extended window: decreased echo quality and severe surface fluctuations. The more prominent either factor is, the larger the product and the greater the extension.

[0081] This indicates a shortened window branch, which is a mirror image of the extended window branch. This occurs when the echo quality is better than the reference and the liquid level fluctuation is lower than the reference. At this point, the calculated value is used as the adjustment coefficient to shorten the window. When the calculated value is greater than or equal to 1, the adjustment coefficient is 1, and the window length remains unchanged.

[0082] This indicates that the window branch should be kept. When the adjustment direction is to keep the window from being adjusted, the adjustment coefficient is set to 1.

[0083] In another specific embodiment, after generating the sampling window length adjustment instruction for the next measurement time window, a dynamic adjustment step is further included to prevent drastic changes in window length between adjacent time windows. The dynamic adjustment step is as follows: calculating the absolute difference between the sampling window length of the next measurement time window and the length value of the current measurement time window.

[0084] When the absolute difference is greater than the preset maximum step size for a single change (the default value is 1 second), if the sampling window length of the next measurement time window is greater than the length of the current measurement time window, the sampling window length of the next measurement time window will be adjusted to the sum of the length of the current measurement time window and the maximum step size for a single change.

[0085] If the sampling window length of the next measurement time window is less than the length of the current measurement time window, then the sampling window length of the next measurement time window is adjusted to the difference between the length of the current measurement time window and the single maximum change step size.

[0086] If the absolute difference is less than or equal to the maximum step size of a single change, the sampling window length of the next measurement time window will not be adjusted.

[0087] Based on the adjusted sampling window length of the next measurement time window, a sampling window length adjustment instruction for the next measurement time window is generated.

[0088] This step generates a sampling window length adjustment instruction based on the combined judgment result of fluctuation level and change trend. During the adjustment process, the current measurement dispersion and echo quality factor are introduced to correct the adjustment coefficient, thus overcoming the contradiction that a single fixed window length cannot simultaneously take into account response speed and stability.

[0089] Secondly, during the window length adjustment process, by constraining the step size of the length change between adjacent measurement time windows, drastic changes in window length caused by misjudgment of fluctuation level or abnormal echo quality are avoided, ensuring a smooth transition in the measurement parameter adjustment process, and further improving the control reliability and measurement result continuity under complex working conditions.

[0090] S6. Perform radar detection within the adjusted measurement time window to output the corresponding liquid level measurement results.

[0091] It should be noted that within the adjusted measurement time window, the radar level gauge performs radar detection according to the adjusted sampling window length, acquiring multiple detection results within that window to form a new sequence of original measurement values. This sequence of original measurement values ​​is then processed to eliminate random errors and obtain the level measurement result corresponding to that measurement time window. For example, the arithmetic mean of multiple original measurement values ​​in the sequence can be calculated after median filtering, and the result is output as the final level measurement result. Simultaneously, the sequence of original measurement values ​​for that measurement time window, along with its range, change per unit time, fluctuation level, and trend, is stored as historical data for subsequent adaptive adjustments to the measurement time window.

[0092] The embodiments of the present invention, through the above-mentioned dynamic window adjustment, can accommodate the different requirements of two working conditions for measurement parameters within the same measurement cycle, thereby taking into account both measurement response speed and measurement accuracy under dynamic working conditions.

[0093] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A liquid level measurement method based on millimeter-wave radar, characterized in that, The method includes: Multiple radar detections are performed within consecutive measurement time windows to obtain a sequence of raw measurement values ​​composed of multiple detection results within each measurement time window; Extract the range and change per unit time from the original measurement sequence to determine the liquid level fluctuation level corresponding to the current measurement time window; The fluctuation trend of liquid surface fluctuation levels is analyzed for multiple consecutive measurement time windows to obtain the changing trend of liquid surface fluctuation state. The changing trend is one of the following: tending to stabilize, fluctuation intensification, and remaining unchanged. The direction for adjusting the window length is determined based on the level of liquid level fluctuation within the current measurement time window and the observed trend. Based on the window length adjustment direction, generate the sampling window length adjustment command for the next measurement time window, and execute the measurement time window adjustment; Radar detection is performed within the adjusted measurement time window to output the corresponding liquid level measurement results; The specific process for determining the level of liquid level fluctuation is as follows: Extract the range and the change per unit time from the original measurement value sequence corresponding to the current measurement time window to form the current fluctuation characteristic pair; Obtain the range and change per unit time of the N consecutive historical measurement time windows preceding the current measurement time window, and construct a set of historical fluctuation feature pairs; The range and change per unit time in the current fluctuation feature pair are compared one by one with the range and change per unit time of each historical fluctuation feature pair in the historical fluctuation feature pair set. The number of historical fluctuation feature pairs in the historical fluctuation feature pair set whose range is less than the current range and whose change per unit time is less than the current change is counted to obtain the first count value. Calculate the ratio of the first count value to the total number of historical fluctuation feature pairs in the set of historical fluctuation feature pairs to obtain the current fluctuation relative index; The level of liquid level fluctuation corresponding to the current measurement time window is determined based on the numerical range of the current fluctuation relative index. The specific steps for the fluctuation trend analysis are as follows: Obtain the liquid level fluctuation levels of multiple consecutive measurement time windows preceding the current measurement time window, and construct a fluctuation level sequence according to the chronological order; Based on the fluctuation level sequence, change point detection is performed, and the direction of fluctuation level change after the most recent change point is determined according to the change point detection results. When the direction of the fluctuation level after the most recent point of change is a continuous increase, the trend of the liquid surface fluctuation state is judged to be an increase in fluctuation. When the direction of the fluctuation level after the most recent point of change is a continuous decrease, the trend of the liquid surface fluctuation state is judged to be stabilizing. When no change point is detected or the direction of change of the fluctuation level after the most recent change point remains unchanged, the trend of the change of the liquid surface fluctuation state is judged to remain unchanged. The process for determining the adjustment direction is as follows: Obtain the liquid level fluctuation level for the current measurement time window, which includes a stable level, a moderate fluctuation level, and a severe fluctuation level. When the liquid level fluctuation tends to stabilize, if the liquid level fluctuation level is stable or moderate, the adjustment direction is to extend the window; if it is severe fluctuation, the window is maintained. When the fluctuation of the liquid surface becomes more intense, the adjustment direction is to shorten the window. When the liquid level fluctuation remains unchanged, the adjustment direction is to maintain the window. The instruction for adjusting the sampling window length to generate the next measurement time window specifically includes: Obtain the standard deviation of multiple raw measurement values ​​within the current measurement time window as the current measurement dispersion; The adjustment coefficient is determined based on the current measurement dispersion and the adjustment direction of the window length, and the product of the current measurement time window length and the adjustment coefficient is used as the target window length. When the adjustment direction is to extend the window, the smaller of the target window length and the preset maximum window length is used as the sampling window length of the next measurement time window; When the adjustment direction is to shorten the window, the larger of the target window length and the preset minimum window length is used as the sampling window length of the next measurement time window; When the orientation is adjusted to hold the window, the target window length is used as the sampling window length for the next measurement time window; Based on the sampling window length of the next measurement time window, generate a sampling window length adjustment instruction for the next measurement time window.

2. The liquid level measurement method based on millimeter-wave radar as described in claim 1, characterized in that: The specific process of the change point detection includes: The volatility levels are converted into volatility scores, resulting in a volatility score sequence; The sliding window is sequentially moved from the start position to the end position of the fluctuation score sequence according to the preset sliding window. At each sliding position, the first mean of the fluctuation score within the sliding window and the second mean of the fluctuation score outside the sliding window are calculated. Calculate the absolute difference between the first mean and the second mean, and mark the sliding position where the absolute difference reaches its maximum value as a candidate change point; The absolute difference at the candidate change point is compared with a preset difference threshold. When the absolute difference is greater than the preset difference threshold, the candidate change point is output as the detected change point. If no candidate change point is detected within the entire length of the fluctuation score sequence that satisfies an absolute difference greater than a preset difference threshold, then when the sequence length of the fluctuation score sequence is greater than a preset length threshold, the starting point or ending point of the fluctuation score sequence will be marked as a change point.

3. The liquid level measurement method based on millimeter-wave radar as described in claim 1, characterized in that: Before determining the adjustment factors, an echo quality check is performed. The specific check steps are as follows: Obtain the peak sidelobe ratio of each frame of radar echo within the current measurement time window, and calculate its average value as the current echo quality factor; If the current echo quality factor is less than the preset echo quality warning threshold, the adjustment direction will be corrected to maintain the window, and the adjustment coefficient will be set to 1. Conversely, if the original adjustment direction is maintained, the echo quality verification will pass.

4. The liquid level measurement method based on millimeter-wave radar as described in claim 3, characterized in that: The specific calculation process for the adjustment coefficient is as follows: Let the current echo quality factor be denoted as Simultaneously, the ratio of the current measurement dispersion to the preset benchmark dispersion is calculated to obtain the current dispersion fluctuation factor. ; Calculate the adjustment factor , , To set the reference echo quality factor, , and Each sets conditions. This indicates that the direction of adjustment is to extend the window. This indicates that the adjustment direction is to shorten the window. This indicates that adjusting the orientation will keep the window open. This represents the function that takes the maximum value. This represents the function that takes the minimum value.

5. The liquid level measurement method based on millimeter-wave radar as described in claim 4, characterized in that: The preset baseline dispersion is updated in the following way: The target window is the measurement time window in which the liquid level fluctuation is stable and the trend remains unchanged. For each target window, the standard deviation of multiple original measurements within the window is calculated as a sample of stationary standard deviations, and the mean of each sample of stationary standard deviations is used as the updated preset benchmark dispersion.

6. The liquid level measurement method based on millimeter-wave radar as described in claim 1, characterized in that: After generating the sampling window length adjustment instruction for the next measurement time window, the following dynamic adjustment steps are also included: Calculate the absolute difference between the sampling window length of the next measurement time window and the length of the current measurement time window; When the absolute difference is greater than the preset maximum step size for a single change, if the sampling window length of the next measurement time window is greater than the length of the current measurement time window, then the sampling window length of the next measurement time window is adjusted to the sum of the length of the current measurement time window and the maximum step size for a single change. If the sampling window length of the next measurement time window is less than the length of the current measurement time window, then the sampling window length of the next measurement time window is adjusted to the difference between the length of the current measurement time window and the single maximum change step size. Based on the adjusted sampling window length of the next measurement time window, a sampling window length adjustment instruction for the next measurement time window is generated.

Images