A pressure intelligent compensation adjusting method and system for a multi-stage adjustable atomizing nozzle

Abstract

The application discloses a pressure intelligent compensation adjusting method and system of a multi-stage adjustable atomizing nozzle, relates to the technical field of atomizing nozzles, and comprises the following steps: collecting pressure signal sets of the nozzle under different working conditions, environment parameters in which the nozzle is located and atomizing images; preprocessing the pressure signal sets and the environment parameters to respectively obtain a plurality of pressure fluctuation basic modes and environment compensation basic modes; performing pressure fluctuation basic mode determination analysis on the nozzle based on the plurality of pressure fluctuation basic modes and the atomizing images, and screening to obtain pressure effective modes; screening to obtain environment compensation effective modes based on the environment compensation basic modes and the pressure signal sets under the current working condition; constructing a mode similarity matrix based on the pressure effective modes and the environment compensation effective modes, and matching to obtain an optimal mode pair, and generating a nozzle pressure compensation adjusting strategy under the current working condition based on the optimal mode pair; and the application improves the atomizing effect and reduces the operation energy consumption.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of atomizing nozzles, in particular to a pressure intelligent compensation adjustment method and system for a multi-stage adjustable atomizing nozzle. BACKGROUND

[0002] At present, as the core component of fluid atomization, the atomization effect of the atomizing nozzle directly affects the operation efficiency and quality. Key indicators of the atomization effect include droplet size distribution, spray cone angle stability, flow uniformity, etc., and these indicators are closely related to the dynamic characteristics of the internal pressure of the nozzle. However, in actual application, the pressure of the nozzle is easily disturbed by multiple factors, leading to fluctuations in the atomization effect. The existing technology has the following significant limitations:

[0003] On the one hand, the existing system mainly uses a single pressure sensor to collect local steady-state pressure, which cannot cover the full-band pressure characteristics of the atomization process. For example, missing low-frequency basic pressure drift, which is easy to cause the output fluctuation of the pump group to cause the spray cone angle to deviate and cannot be perceived. This one-sided monitoring leads to a lag in the prediction of atomization abnormalities, which is often dealt with passively after the quality exceeds the standard, affecting the atomization quality.

[0004] On the other hand, the coupling relationship between pressure fluctuations and environmental parameters (such as temperature, humidity, dust, etc.) has not been modeled, i.e., only through single pressure feedback adjustment, it is difficult to cope with multi-factor coupling scenarios. For example, when the dust is too large, the generated droplet size is too large, the distribution is uneven, leading to a decrease in atomization effect, and large droplets are easy to form liquid accumulation in the pipeline, causing secondary pollution. At the same time, in order to promote the injection of large particle size droplets, higher system pressure is required, and the energy consumption of the pump group and other power equipment increases significantly, increasing energy consumption and operating costs. SUMMARY

[0005] (I) Technical problems solved

[0006] In view of the deficiencies of the prior art, the present application provides a pressure intelligent compensation adjustment method and system for a multi-stage adjustable atomizing nozzle, which obtains correlation analysis results through an initial correlation model and an improved correlation model, and selects pressure fluctuation basic modes that are significantly related to image features as effective modes. Based on environmental parameters, an association function is constructed by combining a pressure signal set to select environmental basic compensation modes that significantly affect the mode as effective modes. Based on the pressure effective modes and the environmental compensation effective modes, the optimal mode pair is obtained, and a pressure compensation adjustment strategy is generated, solving the problems raised in the background art.

[0007] (II) Technical solutions

[0008] To achieve the above purpose, the present application is implemented by the following technical solutions:

[0009] In a first aspect, the application provides a pressure intelligent compensation adjustment method and system for a multi-stage adjustable atomizing nozzle, the method comprising:

[0010] Collecting a pressure signal set of the nozzle, environmental parameters of the nozzle, and an atomization image under different working conditions;

[0011] Pretreating the pressure signal set and the environmental parameters to obtain a plurality of pressure fluctuation basis modes and environmental compensation basis modes, respectively;

[0012] Based on the plurality of pressure fluctuation basis modes and the atomization image, determining and analyzing the pressure fluctuation basis modes of the nozzle to screen a pressure effective mode; based on the environmental compensation basis modes and the pressure signal set under the current working condition, building a rule engine to determine a plurality of influence modes, and based on the plurality of influence modes, screening an environmental compensation effective mode;

[0013] Based on the pressure effective mode and the environmental compensation effective mode, constructing a mode similarity matrix, taking the mode similarity matrix as a cost matrix, matching to obtain an optimal mode pair, and based on the optimal mode pair, generating a nozzle pressure compensation adjustment strategy under the current working condition.

[0014] Further, the pressure signal set includes gas and water flow signals, water supply pressure signals, and gas supply pressure signals, and the environmental parameters include temperature and humidity data and dust concentration data.

[0015] Further, the pretreatment of the pressure signal set includes:

[0016] The pressure signal set is preliminarily decomposed by using a variational mode decomposition algorithm to obtain a plurality of intrinsic mode functions;

[0017] Based on the intrinsic mode functions, the number of extreme points and the standard deviation are extracted, and a characteristic trajectory curve is drawn; based on the points where the derivative value of the characteristic trajectory curve is 0 as the division points, the single monotonic intervals are divided;

[0018] The single monotonic intervals are subjected to hybrid clustering, and the hybrid clustering includes spherical clustering and rectangular clustering, and the extreme points of each single monotonic interval are extracted as the initial spherical center set of the spherical clustering; a monitoring radius of the spherical clustering is preset, the spherical clustering of the characteristic points in the single monotonic interval is performed, and the abnormal characteristic points exceeding the monitoring radius are screened out; the x-axis interval and the y-axis interval of the rectangular clustering are preset, the abnormal characteristic points screened out by the spherical clustering are subjected to rectangular clustering, if the abnormal characteristic points fall into the rectangular interval, a plurality of pressure fluctuation basis modes are divided, otherwise, if the abnormal characteristic points do not fall into any rectangular interval, an unclassified mode is divided.

[0019] Further, based on the plurality of pressure fluctuation basis modes and the atomization image, determining and analyzing the pressure fluctuation basis modes of the nozzle to screen a pressure effective mode, including:

[0020] An initial correlation model is constructed in advance, and the initial correlation model adopts an XGBoost algorithm, the pressure fluctuation basic mode and the image feature are introduced into the initial correlation model, a new composite feature is constructed through feature cross and combination, and key features are screened out through secondary feature screening analysis;

[0021] An adaptive training layer is embedded in the initial correlation model, the screened key features and original features are fused to form an optimized training data set, and random samples are selected from the collected pressure signal set and atomization images to provide iterative training and parameter optimization of the adaptive training layer, iterative update of model parameters, and an improved correlation model is obtained;

[0022] The pressure signal set and the atomization image obtained in real time are layered according to time sequence data, the pressure fluctuation basic mode and the image feature under each time sequence are introduced into the improved correlation model, and it is judged that the features extracted under each time sequence belong to key features.

[0023] Further, the environmental parameters are preprocessed, including:

[0024] An environmental basic value is obtained based on the environmental parameters;

[0025] The environmental compensation basic value is compared with a preset compensation threshold interval to divide a plurality of discrete intervals, and each interval corresponds to a compensation mode:

[0026] If the environmental compensation basic value is less than the preset compensation threshold interval, it is determined as a weak compensation mode;

[0027] If the environmental compensation basic value is in the preset compensation threshold interval, it is determined as a moderate compensation mode;

[0028] If the environmental compensation basic value is greater than the preset compensation threshold interval, it is determined as a strong compensation mode.

[0029] Further, based on the environmental compensation basic mode and the pressure signal set under the current working condition, a rule engine is built to determine a plurality of influence modes, and an environmental compensation effective mode is obtained based on the plurality of influence modes, including:

[0030] Based on the pressure signal set, a five-dimensional monitoring vector is constructed, including gas and water flow, water supply pressure, gas supply pressure, environmental basic value and time stamp;

[0031] extracting rising edge slope features and falling edge slope features of the five-dimensional monitoring vector within a preset time period; constructing a plurality of change vectors based on the rising edge slope and the falling edge slope, constructing an association function between each change vector and an environmental compensation base value, determining an influence mode of the environmental parameter on the pressure signal set based on the association function; at the same time, obtaining a determination coefficient of each association function, sequentially sorting each association function based on the determination coefficient, and determining the most significant influence mode to screen an effective mode of environmental compensation.

[0032] Further, the association function is: taking each change vector as an input feature and the environmental compensation base value as an output target, and using machine learning to establish a regression model of the input feature and the output target.

[0033] Further, taking the mode similarity matrix as a cost matrix, the optimal mode pair is obtained by matching, including:

[0034] Based on the effective pressure mode, the amplitude decay rate and the frequency offset are extracted and combined into a pressure feature vector;

[0035] Based on the effective environmental compensation mode, the goodness of fit and the timestamp are extracted and combined into an environmental compensation feature vector;

[0036] The cosine similarity is used to calculate the similarity of the pressure feature vector and the environmental compensation feature vector to form a mode similarity matrix, which is converted into a cost matrix; the KM algorithm is used to obtain the optimal mode pair with the minimum cost matrix as the target.

[0037] Further, based on the optimal mode pair, a nozzle pressure compensation adjustment strategy under the current working condition is generated, including:

[0038] determining whether the optimal mode pair meets the first adjustment constraint to obtain a first adjustment result; wherein the first adjustment constraint includes a pressure fluctuation constraint;

[0039] If the first adjustment result meets the first adjustment constraint, it is determined whether the optimal mode pair meets the second adjustment constraint to obtain a second adjustment result; otherwise, a second compensation result is obtained, and the second compensation result is that the optimal mode pair needs to be optimized.

[0040] Wherein, the second adjustment constraint includes an atomization effect constraint;

[0041] If the second adjustment result meets the second adjustment constraint, a first compensation result is obtained, and the first compensation result is that the pressure compensation adjustment is successful.

[0042] In a second aspect, the present application provides a pressure intelligent compensation adjustment system for a multi-stage adjustable atomizing nozzle, the system comprising:

[0043] The acquisition module is configured to acquire a pressure signal set of the nozzle, environmental parameters of the nozzle, and an atomization image under different working conditions.

[0044] a preprocessing module for preprocessing the pressure signal set and the environmental parameters to obtain a plurality of pressure fluctuation basis modes and environmental compensation basis modes, respectively;

[0045] a screening analysis module for pressure fluctuation basis mode determination and analysis of the nozzle based on the plurality of pressure fluctuation basis modes and the atomization image, screening to obtain a pressure effective mode; based on the environmental compensation basis mode and the pressure signal set under the current working condition, a rule engine is built to determine a plurality of influence modes, and based on the plurality of influence modes, an environmental compensation effective mode is screened;

[0046] a matching compensation module for constructing a mode similarity matrix based on the pressure effective mode and the environmental compensation effective mode, taking the mode similarity matrix as a cost matrix, matching to obtain an optimal mode pair, and generating a nozzle pressure compensation adjustment strategy under the current working condition based on the optimal mode pair.

[0047] (Three) beneficial effects

[0048] The present application provides a multi-stage adjustable atomizing nozzle pressure intelligent compensation adjustment method and system, which has the following beneficial effects:

[0049] 1. The present application analyzes the potential correlation between pressure fluctuation and atomization image through initial correlation model and improved correlation model, and excavates the key factors affecting atomization quality, providing data-driven decision basis for multi-stage adjustable atomizing nozzle pressure intelligent compensation adjustment;

[0050] 2. The present application screens out pressure fluctuation basis modes significantly related to image features as effective modes based on correlation analysis results; based on environmental parameters, an association function is constructed combining pressure signal set to screen out environmental basis compensation modes significantly affecting modes as effective modes, and the optimal matching pair obtained on this basis can accurately reflect the internal correlation between pressure fluctuation and environmental compensation, so that the pressure compensation adjustment strategy generated based thereon is more in line with actual demand, thereby greatly improving the accuracy of adjustment; at the same time, considering environmental parameters and constructing an association function to screen environmental compensation effective modes makes the system fully consider the influence of different environmental factors on pressure, improving the adaptability and robustness of the system;

[0051] 3. The present application obtains different change vectors through different combination methods, which can reflect the dynamic response characteristics of the nozzle system under different environmental conditions; through single monotonic interval mixed clustering, including using spherical clustering and rectangular clustering, twice screening, and finally through cross combination of X-axis interval type and Y-axis interval type, different pressure fluctuation basis modes are divided, providing accurate signal basis for subsequent mode matching and pressure compensation adjustment strategy generation of the system, ensuring that the pressure compensation adjustment strategy is highly targeted and efficient in response. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 is a flowchart of a pressure intelligent compensation adjustment method according to an example embodiment;

[0053] Figure 2 is a module diagram of a pressure intelligent compensation adjustment system according to an example embodiment. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0055] Embodiment 1:

[0056] The embodiments of the present application provide a pressure intelligent compensation adjustment method and system for a multi-stage adjustable atomizing nozzle; Figure 1 is a flowchart of a pressure intelligent compensation adjustment method according to an example embodiment; please refer to Figure 1 The method comprises the following steps:

[0057] A plurality of sensors are used to synchronously collect a pressure signal set of the nozzle under different working conditions, environmental parameters of the nozzle and atomization images. The pressure signal set includes but is not limited to air-water flow signals, water supply pressure signals and air supply pressure signals. The environmental parameters include temperature and humidity data and dust concentration data.

[0058] Air-water flow signals are collected by air-water flow sensors. A multi-stage sensing array is used to synchronously collect nozzle inlet pressure, outlet pressure and cavity dynamic pressure, to obtain water supply pressure signals and air supply pressure signals, and to generate a full-band pressure signal set. The multi-stage sensing array comprises a plurality of pressure sensors, including an inlet pressure sensor, an outlet pressure sensor and a cavity dynamic pressure sensor array. A multi-channel synchronous data acquisition card (such as NI PXIe-6368) is used to ensure that the sampling clock of air-water flow, inlet pressure, outlet pressure and cavity pressure is unified. Environmental temperature, environmental humidity and dust concentration are collected by temperature sensors, humidity sensors and dust concentration sensors respectively. Atomization images are captured by high-frequency cameras. The above sensors or devices are not drawn in the figure and are adaptively installed. In addition, the time deviation of the sensors can be corrected by an interpolation algorithm (such as linear interpolation) to strictly align all signals in the time dimension and realize synchronous collection.

[0059] The pressure signal set is preprocessed to obtain a plurality of pressure fluctuation basis modes; based on the plurality of pressure fluctuation basis modes and the atomization image, pressure fluctuation basis mode determination analysis of the nozzle is performed to obtain pressure effective modes; the amplitude decay rate and the frequency offset are extracted and combined into a pressure feature vector;

[0060] The pressure signal set is preprocessed, including:

[0061] The pressure signal set is preliminarily decomposed by using a variational mode decomposition (VMD) algorithm to obtain a plurality of intrinsic mode functions (IMFs), and each intrinsic mode function IMF represents a fluctuation component of the pressure signal at different frequencies and scales;

[0062] It should be noted that during the decomposition process, key parameters of VMD need to be set, such as: mode number, quadratic penalty factor / bandwidth control parameter; the mode number can be determined by prior knowledge (for example: the number of main oscillation modes is estimated according to the physical system (such as valve action frequency) generating the pressure signal set); the quadratic penalty factor / bandwidth control parameter affects the bandwidth of each IMF: the larger the value of the quadratic penalty factor, the narrower the bandwidth of the IMF (the higher the frequency resolution), the purer the mode, but the more sensitive to noise, which may lead to mode splitting; the smaller the quadratic penalty factor, the wider the bandwidth of the IMF (the higher the time resolution), the stronger the anti-noise ability, but it may lead to mode blurring or aliasing; the balance between signal reconstruction accuracy and mode bandwidth is determined by adjusting the quadratic penalty factor and determining the mode number; a reasonable mode number value ensures that the decomposition result can fully reflect the signal characteristics and avoid excessive decomposition leading to redundancy;

[0063] During the iteration process, the VMD algorithm continuously updates the center frequency and bandwidth of each mode, so that the energy of each mode is concentrated in a specific frequency range, and effective separation of different frequency components is achieved; each separated IMF component represents a pressure fluctuation basis mode, corresponding to the fluctuation characteristics of the pressure signal at different frequencies and scales, such as low-frequency system inherent vibration and high-frequency cavitation effect fluctuation;

[0064] Finally, the modes obtained by preliminary decomposition are post-processed, the instantaneous frequency and amplitude information of each mode are obtained by Hilbert transform, the physical meaning of the mode is verified, and it is ensured that the separated basis mode can truly reflect the pressure fluctuation characteristics, laying a foundation for subsequent analysis combined with the atomization image; the decomposition technology of the variational mode decomposition (VMD) algorithm belongs to a conventional technology, which will not be described here;

[0065] Based on the intrinsic mode function IMF, the number of extreme points and the standard deviation are extracted, the number of extreme points is taken as the horizontal axis, the standard deviation is taken as the vertical axis, these points are connected according to the time sequence of the intrinsic mode function IMF, and a characteristic trajectory curve is drawn; based on the points with a derivative value of 0 of the characteristic trajectory curve, a plurality of monotonic intervals are divided.

[0066] The monotonic intervals are mixed clustered, and the mixed clustering includes spherical clustering and rectangular clustering, and extreme points (such as peak points and valley points) of each monotonic interval are extracted as an initial spherical center set of the spherical clustering;

[0067] A monitoring radius of the spherical clustering is preset, the spherical clustering is performed on feature points in the monotonic interval, and an abnormal feature point exceeding the monitoring radius is screened out; an x-axis interval and a y-axis interval of the rectangular clustering are preset, the rectangular clustering is performed on the abnormal feature point screened out by the spherical clustering, if the abnormal feature point falls into a rectangular interval, a plurality of pressure fluctuation basic modes are divided, otherwise, if the abnormal feature point does not fall into any rectangular interval, an unclassified mode is divided;

[0068] If the abnormal feature point falls into the rectangular interval, the plurality of pressure fluctuation basic modes are divided, including:

[0069] A preset interval type is included, an x-axis interval type includes a low-frequency interval, a medium-frequency interval and a high-frequency interval, and the low-frequency interval is smaller than the medium-frequency interval which is smaller than the high-frequency interval; a y-axis interval type includes a small-amplitude interval, a medium-amplitude interval and a large-amplitude interval, and the small-amplitude interval is smaller than the medium-amplitude interval which is smaller than the large-amplitude interval;

[0070] Different pressure fluctuation basic modes are divided by cross combination of the x-axis interval type and the y-axis interval type:

[0071]

[0072] For example, the low-frequency violent fluctuation mode represents high-frequency small-amplitude fluctuation, for example, the x-axis interval is [30, 50] times per second, and the y-axis interval is [0.1, 0.5] MPa; the high-frequency micro-vibration fluctuation mode represents low-frequency large-amplitude fluctuation, for example, the x-axis interval is [5, 15] times per second, and the y-axis interval is [1.0, 2.0] MPa;

[0073] It should be noted that in the process of mixed clustering, the monitoring radius of the spherical clustering, the number of radius intervals and the x-axis interval and the y-axis interval range of the rectangular clustering are adaptively adjusted based on the statistical characteristics of the feature point distribution, and specifically include: based on the statistical characteristics of the historical pressure data, the mean value and the standard deviation of the number of extreme points and the standard deviation are calculated, and then:

[0074] The x-axis interval = [Q1(X)-k*IQR(X), Q3(X)+k*IQR(X)];

[0075] The y-axis interval = [Q1(Y)-m*IQR(Y), Q3(Y)+m*IQR(Y)]; wherein Q1 is a lower quartile, Q3 is an upper quartile, IQR is a quartile, and k and m are adjustable coefficients (1≤k, m≤3);

[0076] Based on several pressure fluctuation basis modes and atomization images, the pressure fluctuation basis mode of the nozzle is determined and analyzed, and the effective pressure mode is obtained by screening, including:

[0077] Based on the atomization image, the image features are extracted by using feature extraction technology; wherein the feature extraction technology includes but is not limited to gray level co-occurrence matrix (GLCM), local binary pattern (LBP), and histogram of oriented gradients (HOG) algorithm, and the extracted image features include but are not limited to droplet size distribution, spray cone angle, texture roughness and edge intensity parameters, and an image feature matrix is constructed;

[0078] An initial correlation model is constructed in advance, and the initial correlation model adopts XGBoost algorithm, the pressure fluctuation basis mode and the image features are introduced into the initial correlation model, and a new composite feature is constructed by feature cross and combination; the key features are selected by using secondary feature screening analysis;

[0079] Primary feature screening: an association matrix between the pressure fluctuation basis mode (such as frequency, amplitude) and the image features is established, and the correlation between the two is quantified by calculating the Pearson correlation coefficient; for example, if the frequency of a certain pressure mode is highly correlated with the droplet size fluctuation frequency (such as Pearson correlation coefficient > 0.8), it indicates that this mode has an important influence on the droplet size change; for example: the pressure fluctuation under different working conditions has different effects on the nozzle atomization characteristics, assuming that the gas supply pressure is the research object, assuming that the nozzle is selected as a non-symmetrical inlet structure, the spatial distribution of atomized droplets is affected by the non-symmetrical inlet structure of the nozzle, and presents a non-symmetrical distribution in the drum cavity, the droplet size changes significantly with the spatial position, the droplet size is the largest outside the atomization range, and the droplet size is the smallest inside the atomization space, with the increase of the gas supply pressure, the average droplet size decreases, and with the increase of the gas-liquid relative velocity, the maximum droplet size and the average droplet size both decrease, and the atomization effect is significantly improved;

[0080] According to the mechanism of nozzle atomization, the process of liquid atomization is mainly divided into three stages:

[0081] The first stage, the liquid and the gas are respectively introduced into the nozzle structure through the inlet, and are guided to the nozzle inlet;

[0082] The second stage, after being sprayed, the liquid undergoes a primary atomization, at this time the liquid forms a large area liquid film at the nozzle outlet, and further breaks into large diameter droplets;

[0083] The third stage, the primary atomized droplets continue to deform and break under the action of the gas flow, and the droplet diameter is further reduced, which is the secondary atomization process;

[0084] Secondary feature selection: A feature selection method combining recursive feature elimination (RFE) and variance inflation factor (VIF) is used to select key features with strong correlation and low redundancy; for modes with weak correlation, further evaluation and selection are carried out in combination with mode confidence criteria (such as mathematical fit, physical correlation, algorithm robustness, etc.).

[0085] An adaptive training layer is embedded in the initial association model. The selected key features are fused with the original features to form an optimized training dataset. Samples are randomly selected from the collected pressure signal set and fogging images (e.g., the training set and validation set are randomly divided in a 7:3 ratio). The adaptive training layer is iteratively trained and the parameters are tuned. The importance of each feature to the fogging effect is quantitatively evaluated (by calculating the gain rate and number of splits of the features in the decision tree). The model parameters are iteratively updated to obtain an improved association model with higher accuracy.

[0086] Meanwhile, the obtained pressure signal set and atomized image are layered according to time series data, and the basic modes of pressure fluctuation and image features under each time series are imported into the improved correlation model to determine that the features extracted under each time series are key features.

[0087] It should be noted that during the model calculation process, the image features and pressure feature vectors involved need to be Z-score standardized to eliminate the influence of dimensions.

[0088] The above process analyzes the potential correlation between pressure fluctuations and atomization images through initial and improved correlation models, uncovers key factors affecting atomization quality, and provides data-driven decision-making basis for intelligent pressure compensation adjustment of multi-level adjustable atomizing nozzles. Based on the correlation analysis results, pressure modes that are significantly related to image features are selected as effective modes, providing a precise signal basis for subsequent mode matching (e.g., KM algorithm matching environmental compensation modes) and pressure compensation adjustment strategy generation (e.g., adaptive PID parameter optimization), ensuring that the adjustment strategy is highly targeted and has a high response efficiency.

[0089] Environmental parameters are preprocessed to obtain several basic environmental compensation modes. Based on the basic environmental compensation modes and the pressure signal set under the current working conditions, a rule engine is built to determine several influencing modes. Based on the influencing modes, effective environmental compensation modes are obtained by screening, and the goodness of fit and timestamps are extracted and combined into an environmental compensation feature vector.

[0090] Environmental parameter preprocessing includes:

[0091] By analyzing the transformation of environmental parameters through a nonlinear mapping function, the parameters are weighted and combined and then subjected to nonlinear transformation to obtain the basic value of environmental compensation, which takes the value range of [0, 1]. The larger the value, the stronger the compensation required.

[0092] The environmental compensation base value is compared with a preset compensation threshold interval to divide a plurality of discrete intervals, each interval corresponding to a compensation mode:

[0093] If the environmental compensation base value is less than the preset compensation threshold interval, it is determined to be a weak compensation mode, which requires less compensation;

[0094] If the environmental compensation base value is in the preset compensation threshold interval, it is determined to be a moderate compensation mode, which requires moderate compensation;

[0095] If the environmental compensation base value is greater than the preset compensation threshold interval, it is determined to be a strong compensation mode, which requires greater compensation;

[0096] It should be noted that in the present embodiment, the preset compensation threshold interval is obtained based on statistical analysis of historical environmental compensation base values, and the mean and standard deviation are calculated based on the historical environmental compensation base values. The compensation threshold interval is set as a combination of the mean and the multiple of the standard deviation, for example: the minimum value of the compensation threshold interval is set as the mean minus 2 times the standard deviation, and the maximum value of the compensation threshold interval is set as the mean plus 2 times the standard deviation;

[0097] Based on the environmental compensation base mode and the pressure signal set under the current working condition, a rule engine is built to determine a plurality of influence modes, and based on the plurality of influence modes, an effective environmental compensation mode is obtained, including:

[0098] Based on the pressure signal set, a five-dimensional monitoring vector is constructed, including gas and water flow, water supply pressure, gas supply pressure, environmental base value and time stamp;

[0099] Within a preset time period, the rising slope feature and the falling slope feature of the five-dimensional monitoring vector are extracted;

[0100] For the time series data of each dimension in the five-dimensional monitoring vector, the slope in each window is calculated using a sliding window, and according to the positive and negative of the slope, it is divided into rising slope and falling slope. The average value of the rising slope and the falling slope is calculated as the corresponding rising slope feature and falling slope feature;

[0101] Based on the rising slope and the falling slope, a plurality of change vectors are constructed, including:

[0102] For example:

[0103] The rising slope and the falling slope corresponding to the gas and water flow are marked as a1 and a2 respectively;

[0104] The rising slope and the falling slope corresponding to the water supply pressure are marked as b1 and b2 respectively;

[0105] The rising edge slope and the falling edge slope corresponding to the air supply pressure are marked as c1 and c2 respectively;

[0106] The compensation mode in which the environmental compensation base value is located is marked as mode;

[0107] The first change vector: [a1+a2, b1+b2, c1+c2, mode]; the second change vector: [a1+a2, b1+b2, c1-c2, mode]; the third change vector: [a1+a2, b1-b2, c1+c2, mode]; the fourth change vector: [a1-a2, b1+b2, c1+c2, mode]; the fifth change vector: [a1+a2, b1-b2, c1-c2, mode]; the sixth change vector: [a1-a2, b1+b2, c1-c2, mode]; the seventh change vector: [a1-a2, b1-b2, c1+c2, mode]; the eighth change vector: [a1-a2, b1-b2, c1-c2, mode];

[0108] Different change vectors are obtained through different combination modes, which can reflect the dynamic response characteristics of the nozzle system under different environmental conditions; for example, [a1+a2, b1+b2, c1+c2] reflects the influence of the overall rising / falling response capability of the system on environmental compensation; and [a1-a2, b1-b2, c1-c2] reflects the influence of the three combinations of rising / falling response differences on environmental compensation;

[0109] An association function between each change vector and the environmental compensation base value is constructed, and based on the association function, the influence mode of the environmental parameter on the pressure signal set is determined;

[0110] Among them, each change vector is used as an input feature, and the environmental compensation base value is used as an output target, a regression model of the input feature and the output target is established using machine learning, and the regression model is the association function; the determination coefficient R 2 of each association function is calculated 2 , and the determination coefficient R 2 is used to evaluate the fitting degree of the association function; based on the determination coefficient R

[0111] It should be noted that the determination coefficient R 2 represents the proportion of the variation of the dependent variable (environmental compensation base value) that can be explained by the independent variable (change vector), and the value range is [0, 1], and the value closer to 1 indicates that the model fitting effect is better;

[0112] First, set an R 2 threshold (such as 0.7), and filter out the R 2the correlation functions whose values are greater than the threshold value are considered to be able to effectively reflect the influence mode of the environmental parameter on the pressure signal set;

[0113] Then, in the screened correlation functions, the curve with the smallest fluctuation and the most stable is selected; for example, the stability of the curve can be evaluated by calculating the variance, standard deviation or mean absolute deviation of the curve, and the smaller the value, the more stable the curve;

[0114] Finally, the screened correlation functions are analyzed for feature importance to determine the contribution of each feature (such as a1, b1, c1) to the environmental compensation base value, so as to determine the environmental compensation effective mode; each effective mode corresponds to a specific combination of change vectors and influence mode; the process of feature importance analysis is a routine technique and will not be described here;

[0115] For example, the correlation functions between each change vector and the environmental compensation base value are constructed, including:

[0116] The correlation function of the first change vector and the environmental compensation base value is constructed, which is determined as the first influence mode of the environmental parameter on the pressure signal set; the correlation function of the second change vector and the environmental compensation base value is constructed, which is determined as the second influence mode of the environmental parameter on the pressure signal set; the correlation function of the third change vector and the environmental compensation base value is constructed, which is determined as the third influence mode of the environmental parameter on the pressure signal set; the correlation function of the fourth change vector and the environmental compensation base value is constructed, which is determined as the fourth influence mode of the environmental parameter on the pressure signal set; the correlation function of the fifth change vector and the environmental compensation base value is constructed, which is determined as the fifth influence mode of the environmental parameter on the pressure signal set; the correlation function of the sixth change vector and the environmental compensation base value is constructed, which is determined as the sixth influence mode of the environmental parameter on the pressure signal set; the correlation function of the seventh change vector and the environmental compensation base value is constructed, which is determined as the seventh influence mode of the environmental parameter on the pressure signal set; the correlation function of the eighth change vector and the environmental compensation base value is constructed, which is determined as the eighth influence mode of the environmental parameter on the pressure signal set;

[0117] Considering the environmental parameters and constructing the correlation functions to screen the environmental compensation effective mode, so that the system can fully consider the influence of different environmental factors on the pressure; through the optimal mode pair of the generated compensation adjustment strategy, the system can better cope with the changes of the environment, so that the system can still maintain stable pressure output under complex and variable environmental conditions, and improve the adaptability and robustness of the system.

[0118] Based on the pressure effective mode and the environmental compensation effective mode, a mode similarity matrix is constructed, the mode similarity matrix is taken as a cost matrix, and an optimal mode pair is matched to obtain a nozzle pressure compensation adjustment strategy under the current working condition;

[0119] The modal similarity matrix is taken as a cost matrix, and an optimal modal pair is obtained by matching, including:

[0120] The pressure feature vector and the environment compensation feature vector are obtained, and the similarity of the pressure feature vector and the environment compensation feature vector is calculated by using cosine similarity: ; in the formula, M p represents the pressure feature vector, M e represents the environment compensation feature vector, p i represents the i-th feature in the pressure feature vector, including but not limited to amplitude decay rate and frequency offset, e j represents the j-th feature in the environment compensation feature vector, including but not limited to goodness of fit and timestamp; and a two-dimensional modal similarity matrix S (a two-dimensional matrix of m*n, S ij represents the similarity of p i and e i , the greater the value, the more similar), and is converted into a cost matrix: C = max (S) - S ij ; that is, the higher the similarity, the lower the cost;

[0121] The KM algorithm is used to find the modal matching combination with the minimum total cost based on the cost matrix, that is, to find the most similar environment compensation effective mode for each pressure effective mode under the premise of ensuring global optimization, and form an optimal modal pair;

[0122] Based on the optimal modal pair, a pressure compensation adjustment strategy is generated, including:

[0123] It is judged whether the optimal modal pair satisfies the first adjustment constraint to obtain a first adjustment result; wherein the first adjustment constraint includes a pressure fluctuation constraint, specifically: a reasonable interval of pressure fluctuation of the pressure signal set is set, and the interval is determined based on the pressure parameter range of stable operation of the equipment, and if the pressure fluctuation value of the pressure effective mode in the optimal modal pair is within the interval, it is considered that the pressure fluctuation constraint is satisfied;

[0124] If the first adjustment result satisfies the first adjustment constraint, it is judged whether the optimal modal pair satisfies the second adjustment constraint to obtain a second adjustment result; otherwise, a second compensation result is obtained, and the second compensation result is the optimal modal pair that needs to be optimized;

[0125] Wherein, the second adjustment constraint includes an atomization effect constraint, specifically, a threshold value of the system set atomization effect is reached, and if the pressure compensation adjustment strategy generated based on the optimal modal pair can make the atomization effect index reach or exceed the threshold value, it is considered that the atomization effect constraint is satisfied;

[0126] If the second adjustment result satisfies the second adjustment constraint, a first compensation result is obtained, and the first compensation result is that the pressure compensation adjustment is successful;

[0127] Specifically, after obtaining the optimal modal pair, the first task is to compare the pressure fluctuation data in the pressure effective modal with the reasonable interval of the pressure fluctuation in the first adjustment constraint; for example, the pressure range of the stable operation of the device is 0.2-0.8 MPa, if the pressure fluctuation of the pressure effective modal in the optimal modal pair is always within this range, then the first adjustment result satisfies the first adjustment constraint; if the pressure fluctuation exceeds the range, such as reaching 0.1 MPa or 0.9 MPa, then the first adjustment result does not satisfy the constraint, and the second compensation result is directly output, that is, it indicates that the optimal modal pair needs to be optimized, and the modal pair may need to be matched again or the characteristics of the existing modal pair need to be adjusted;

[0128] In the process of pressure intelligent compensation adjustment, the adjustment valve opening and the variable frequency pump speed in the actuator are usually controlled based on the pressure compensation adjustment strategy to realize the pressure intelligent compensation adjustment of the multi-stage adjustable atomizing nozzle, for example: the adjustment parameter combination of the multi-stage nozzle assembly (such as valve core opening, spring stiffness) is designed, based on the adjustment parameter combination, the actuator is driven to dynamically reconstruct the internal flow channel structure of the nozzle, and the cooperative optimization of pressure intelligent compensation and atomization effect is realized. This precise pressure compensation adjustment strategy can avoid energy waste caused by improper adjustment (for example: unnecessary energy consumption is reduced by reasonably controlling the adjustment valve opening and the variable frequency pump speed);

[0129] The above entire process from the screening of the effective modal to the acquisition of the optimal matching pair, and then to the generation of the pressure compensation adjustment strategy, forms a complete intelligent analysis and decision-making process; without too much manual intervention, automatic and accurate compensation adjustment of the pressure of the multi-stage adjustable atomizing nozzle can be realized, the error and hysteresis of human operation are reduced, the adjustment efficiency is improved, and intelligent adjustment is realized.

[0130] Embodiment 2:

[0131] The embodiment of the application provides a pressure intelligent compensation adjustment system of a multi-stage adjustable atomizing nozzle; Figure 2 It is a module schematic diagram of the pressure intelligent compensation adjustment system shown according to an example embodiment; please refer to Figure 2 The system comprises a collection module, a preprocessing module, a screening analysis module and a matching compensation module, and the collection module, the preprocessing module, the screening analysis module and the matching compensation module are in communication connection;

[0132] The collection module is used for collecting the pressure signal set of the nozzle under different working conditions, the environmental parameters of the nozzle and the atomization image;

[0133] The preprocessing module is used for preprocessing the pressure signal set and the environmental parameters to obtain a plurality of pressure fluctuation basic modes and environmental compensation basic modes respectively;

[0134] The screening analysis module is used for pressure fluctuation basic mode judgment analysis of the nozzle based on a plurality of pressure fluctuation basic modes and atomization images, and screening to obtain a pressure effective mode; a rule engine is built based on an environmental compensation basic mode and a pressure signal set under a current working condition, a plurality of influence modes are determined, and an environmental compensation effective mode is screened based on the plurality of influence modes;

[0135] The matching compensation module constructs a mode similarity matrix based on the pressure effective mode and the environmental compensation effective mode, takes the mode similarity matrix as a cost matrix, matches to obtain an optimal mode pair, and generates a nozzle pressure compensation adjustment strategy under the current working condition based on the optimal mode pair.

[0136] In the application, the plurality of formulas involved are dimensionless values, and the formulas are obtained by software simulation based on a large amount of collected data to obtain a formula of the nearest real situation, and the formula is set by a person skilled in the art according to the actual situation.

[0137] The above embodiments can be realized wholly or partially by software, hardware, firmware or any other combination. When realized by software, the above embodiments can be realized in the form of a computer program product wholly or partially. Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized by hardware or software depends on the specific application and design constraints of the technical solutions.

[0138] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, and can be located in one place or distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0139] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for intelligent pressure compensation adjustment of a multi-stage adjustable atomizing nozzle, characterized in that, The method includes: Collect pressure signal sets of the nozzle, environmental parameters of the nozzle, and atomization images under different operating conditions; The pressure signal set and the environmental parameters are preprocessed to obtain several pressure fluctuation fundamental modes and environmental compensation fundamental modes, respectively. Based on several pressure fluctuation fundamental modes and the atomization image, the pressure fluctuation fundamental modes of the nozzle are determined and analyzed, and effective pressure modes are obtained by screening. Based on the environmental compensation fundamental modes and the pressure signal set under the current working conditions, a rule engine is built to determine several influence modes, and effective environmental compensation modes are obtained by screening based on several influence modes. A mode similarity matrix is ​​constructed based on the effective pressure mode and the effective environmental compensation mode. The mode similarity matrix is ​​used as the cost matrix to match and obtain the optimal mode pair. Based on the optimal mode pair, a nozzle pressure compensation adjustment strategy under the current working condition is generated. The preprocessing of the pressure signal set includes: using a variational mode decomposition algorithm to initially decompose the pressure signal set to obtain several intrinsic mode functions (IMFs); based on the IMFs, extracting the number of extreme points and standard deviation, and plotting feature trajectory curves; using points where the derivative of the feature trajectory curve is 0 as dividing points to divide the signal set into several monotonic intervals; performing hybrid clustering on the monotonic intervals, including spherical clustering and rectangular clustering, and extracting the extreme points of each monotonic interval as the initial set of sphere centers for the spherical clusters; presetting the monitoring radius of the spherical clusters, and filtering out abnormal feature points exceeding the monitoring radius for the spherical clusters of feature points within the monotonic intervals; presetting the x-axis and y-axis intervals of the rectangular clusters, and for the rectangular clusters of abnormal feature points filtered out by the spherical clusters, if the abnormal feature points fall within the rectangular intervals, they are classified into several basic pressure fluctuation modes; otherwise, if the abnormal feature points do not fall within any rectangular intervals, they are classified as unclassified modes. The preprocessing of the environmental parameters includes: obtaining a baseline environmental value based on the environmental parameters; comparing the baseline environmental compensation value with a preset compensation threshold range to divide the data into multiple discrete ranges, each range corresponding to a compensation mode: if the baseline environmental compensation value is less than the preset compensation threshold range, it is determined to be a weak compensation mode; if the baseline environmental compensation value is within the preset compensation threshold range, it is determined to be a moderate compensation mode; if the baseline environmental compensation value is greater than the preset compensation threshold range, it is determined to be an intense compensation mode.

2. The intelligent pressure compensation and adjustment method for a multi-stage adjustable atomizing nozzle according to claim 1, characterized in that, The pressure signal set includes air and water flow signals, water supply pressure signals, and air supply pressure signals. The environmental parameters include temperature and humidity data and dust concentration data.

3. The intelligent pressure compensation and adjustment method for a multi-stage adjustable atomizing nozzle according to claim 1, characterized in that, The step of analyzing and determining the pressure fluctuation fundamental modes of the nozzle based on several pressure fluctuation fundamental modes and the atomization image, and selecting effective pressure modes, includes: Based on the fogged image, feature extraction techniques are used to extract image features; An initial correlation model is pre-constructed using the XGBoost algorithm. The basic modes of pressure fluctuation and image features are imported into the initial correlation model. New composite features are constructed through feature crossover and combination. Secondary feature screening analysis is used to select key features. An adaptive training layer is embedded in the initial association model. The selected key features are fused with the original features to form an optimized training dataset. Samples are randomly selected from the collected pressure signal set and fogged image. The adaptive training layer is iteratively trained and the parameters are tuned. The model parameters are iteratively updated to obtain an improved association model. The real-time acquired pressure signal set and atomized image are stratified according to time series data. The basic modes of pressure fluctuation and image features under each time series are imported into the improved correlation model to determine whether the features extracted under each time series are key features.

4. The intelligent pressure compensation and adjustment method for a multi-stage adjustable atomizing nozzle according to claim 1, characterized in that, The rule engine is built based on the environmental compensation basic mode and the pressure signal set under the current operating conditions to determine several influence modes, and to obtain effective environmental compensation modes based on the several influence modes, including: Based on the pressure signal set, a five-dimensional monitoring vector is constructed, including gas and water flow rate, water supply pressure, gas supply pressure, environmental baseline value, and timestamp; Within a preset time period, the rising edge slope features and falling edge slope features of the five-dimensional monitoring vector are extracted; several change vectors are constructed based on the rising edge slope and falling edge slope, and a correlation function is constructed between each change vector and the environmental compensation baseline value. Based on the correlation function, the influence mode of environmental parameters on the pressure signal set is determined; at the same time, the determination coefficient of each correlation function is obtained, and the correlation functions are ordered based on the determination coefficient to determine the most significant influence mode, so as to screen the effective mode of environmental compensation.

5. The intelligent pressure compensation and adjustment method for a multi-stage adjustable atomizing nozzle according to claim 4, characterized in that, The correlation function is as follows: taking each change vector as input features and the environmental compensation baseline value as output target, a regression model of input features and output target is established using machine learning.

6. The intelligent pressure compensation and adjustment method for a multi-stage adjustable atomizing nozzle according to claim 1, characterized in that, The step of using the modality similarity matrix as a cost matrix to match and obtain the optimal modality pair includes: Based on the effective pressure mode, the amplitude attenuation rate and frequency offset are extracted and combined into a pressure feature vector; Based on the effective modes of environmental compensation, the goodness of fit and timestamps are extracted and combined into an environmental compensation feature vector; The similarity between the pressure feature vector and the environmental compensation feature vector is calculated using cosine similarity to form a modal similarity matrix, which is then converted into a cost matrix. The KM algorithm is used to obtain the optimal modal pair with the minimum cost matrix as the objective.

7. The intelligent pressure compensation adjustment method for a multi-stage adjustable atomizing nozzle according to claim 1, characterized in that, The nozzle pressure compensation and adjustment strategy based on the optimal mode pair for generating the current operating condition includes: Determine whether the optimal mode pair satisfies the first adjustment constraint to obtain the first adjustment result; wherein, the first adjustment constraint includes the pressure fluctuation constraint; If the first adjustment result satisfies the first adjustment constraint, determine whether the optimal mode pair satisfies the second adjustment constraint and obtain the second adjustment result; otherwise, obtain the second compensation result, and the second compensation result indicates that the optimal mode pair needs to be optimized. The second adjustment constraint includes atomization effect constraints; If the second adjustment result satisfies the second adjustment constraint, the first compensation result is obtained, and the first compensation result indicates that the pressure compensation adjustment is successful.

8. A pressure intelligent compensation and adjustment system for a multi-stage adjustable atomizing nozzle, characterized in that, include: The acquisition module is used to acquire pressure signal sets of the nozzle, environmental parameters of the nozzle, and atomization images under different operating conditions; The preprocessing module is used to preprocess the pressure signal set and the environmental parameters to obtain several pressure fluctuation basic modes and environmental compensation basic modes, respectively. The filtering and analysis module is used to determine and analyze the pressure fluctuation basic mode of the nozzle based on several pressure fluctuation basic modes and the atomization image, and filter to obtain the effective pressure mode; based on the environmental compensation basic mode and the pressure signal set under the current working condition, a rule engine is built to determine several influence modes, and the effective environmental compensation mode is obtained based on several influence modes. The matching compensation module constructs a mode similarity matrix based on the effective pressure mode and the effective environmental compensation mode, uses the mode similarity matrix as the cost matrix, matches to obtain the optimal mode pair, and generates a nozzle pressure compensation adjustment strategy under the current operating condition based on the optimal mode pair. The preprocessing of the pressure signal set includes: using a variational mode decomposition algorithm to initially decompose the pressure signal set to obtain several intrinsic mode functions (IMFs); based on the IMFs, extracting the number of extreme points and standard deviation, and plotting feature trajectory curves; using points where the derivative of the feature trajectory curve is 0 as dividing points to divide the signal set into several monotonic intervals; performing hybrid clustering on the monotonic intervals, including spherical clustering and rectangular clustering, and extracting the extreme points of each monotonic interval as the initial set of sphere centers for the spherical clusters; presetting the monitoring radius of the spherical clusters, and filtering out abnormal feature points exceeding the monitoring radius for the spherical clusters of feature points within the monotonic intervals; presetting the x-axis and y-axis intervals of the rectangular clusters, and for the rectangular clusters of abnormal feature points filtered out by the spherical clusters, if the abnormal feature points fall within the rectangular intervals, they are classified into several basic pressure fluctuation modes; otherwise, if the abnormal feature points do not fall within any rectangular intervals, they are classified as unclassified modes. The preprocessing of the environmental parameters includes: obtaining a baseline environmental value based on the environmental parameters; comparing the baseline environmental compensation value with a preset compensation threshold range to divide the data into multiple discrete ranges, each range corresponding to a compensation mode: if the baseline environmental compensation value is less than the preset compensation threshold range, it is determined to be a weak compensation mode; if the baseline environmental compensation value is within the preset compensation threshold range, it is determined to be a moderate compensation mode; if the baseline environmental compensation value is greater than the preset compensation threshold range, it is determined to be an intense compensation mode.

Images