Intelligent impurity removal control method and system

Abstract

The application discloses a kind of intelligent impurity removal control method and system, it is related to intelligent impurity removal technical field, including, using genetic algorithm to optimize the specific configuration of cleaning parameter in cleaning mode, the optimized cleaning parameter is passed to actuating mechanism and starts cleaning program, continuously monitor operating state in the cleaning process, obtain cleaning effect evaluation report.The application solves the multi-objective conflict of cleaning parameter by defining the multi-objective fitness function of impurity removal efficiency, equipment utilization rate, environmental impact degree, using the global search capability of genetic algorithm.On the basis of guaranteeing impurity removal effect, the comprehensive optimization of reducing reagent consumption and reducing the loss of equipment is realized, the stability of cleaning process is maintained through dynamic fine-tuning mechanism, and clean impurity removal technology is developed towards green and efficient direction.

Description

Technical Field

[0001] This invention relates to the field of intelligent impurity removal technology, and in particular to an intelligent impurity removal control method and system. Background Technology

[0002] With the development of industrial and intelligent manufacturing technologies, the demand for refined impurity treatment has increased significantly. Conventional impurity removal control methods mainly rely on manual experience to set fixed thresholds, and their technical approach can be traced back to the mechanical adjustment architecture based on PID controllers in the 1980s. In complex industrial scenarios, the nonlinear coupling relationship of production parameters and random disturbances severely restrict the flexibility of intelligent impurity removal control systems, making it difficult to achieve full-process adaptive optimization.

[0003] The existing intelligent cleaning control methods have technical bottlenecks in two aspects: the pre-trained reinforcement learning model cannot respond to sudden changes in working conditions in real time, resulting in a delay in the switching of cleaning modes; and there is a lack of dynamic balance mechanism between cleaning efficiency, energy consumption constraints and equipment wear and tear. In the optimization stage of cleaning parameters, grid search or gradient descent methods are often used, which are prone to getting stuck in local optima. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent impurity removal control method that solves the problems of failure of multi-objective collaborative optimization of cleaning parameters and poor adaptability to dynamic environment in intelligent impurity removal control methods.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an intelligent impurity removal control method, which includes collecting operational data and preprocessing it to obtain a standardized dataset;

[0008] We use time series analysis and deep learning models to mine the parameter variation patterns in standardized datasets, and combine them with the Q-learning algorithm to build an adaptive learning model that outputs the optimal parameter combination.

[0009] Based on the optimal parameter combination, the characteristics of impurities are detected using a multimodal sensor, and a deep learning model is used to classify and identify the characteristics of impurities, generating specific impurity type labels. Based on the impurity type, the most suitable cleaning mode is selected from a multi-mode cleaning mode library.

[0010] The specific configuration of cleaning parameters in the cleaning mode is optimized using a genetic algorithm. The optimized cleaning parameters are then passed to the actuator to start the cleaning program. The operating status is continuously monitored during the cleaning process to obtain a cleaning effect evaluation report.

[0011] As a preferred embodiment of the intelligent impurity removal control method of the present invention, the preprocessing includes detecting and removing outliers;

[0012] The preprocessed running data is standardized to generate a standardized dataset.

[0013] As a preferred embodiment of the intelligent impurity removal control method described in this invention, the method includes the following steps: utilizing time series analysis and deep learning models to mine the parameter variation patterns in a standardized dataset.

[0014] By selecting fluid pressure, water flow velocity, and pressure difference across the membrane as input variables and setting timestamp indices, a time series feature data matrix is ​​constructed.

[0015] Input the time series feature data matrix into the LSTM model to generate predicted values ​​for future operating states;

[0016] By comparing the predicted values ​​of future operating status with the time series feature data matrix, the trend patterns of parameter changes in the standardized dataset can be identified.

[0017] As a preferred embodiment of the intelligent impurity removal control method described in this invention, the following steps are included: constructing an adaptive learning model by combining the Q-learning algorithm to output the optimal parameter combination.

[0018] The Q-learning algorithm is used to define the state space, action space, and reward function, and a Q-table is created.

[0019] Input the predicted value of the future running state into the state space, and select the optimal action from the action space of the Q table based on the current actual running state using a greedy strategy.

[0020] Based on the optimal action and the reward function, the adjusted running state and reward value are recorded, the Q value in the Q table is updated, and the loop is iterated until the Q table converges, thus obtaining the optimal parameter combination.

[0021] As a preferred embodiment of the intelligent impurity removal control method of the present invention, the method includes the following steps: detecting impurity characteristics using a multimodal sensor based on the optimal parameter combination, classifying and identifying the impurity characteristics using a deep learning model, and generating specific impurity type labels.

[0022] Using a linear particle size analyzer combined with the principle of laser scattering, the intensity distribution of laser light scattered by particles in the water flow is recorded, and the size range and distribution of particles are statistically analyzed to generate a particle size distribution map.

[0023] The ultrasonic sensor is used to emit high-frequency sound waves to the surface of the water filtration membrane and receive the reflected echo signal to calculate the thickness of the deposits on the surface of the water filtration membrane and generate a deposit thickness distribution map.

[0024] pH and ORP sensors are used to monitor changes in pH and ORP values ​​in water bodies to determine whether the water contains organic matter and biofilm, and a water quality chemical characteristic report is generated based on the results.

[0025] The particle size distribution map, the attachment thickness distribution map, and the water quality chemical characteristics report are combined into a multi-dimensional impurity feature set and converted into image form for input into the CNN model.

[0026] CNN models generate feature maps by sliding convolutional kernels across a multidimensional set of impurity features.

[0027] The ReLU activation function is used to remove negative values ​​from the feature map and retain positive values. At the same time, max pooling is performed to reduce the resolution of the feature map and obtain high-level abstract features.

[0028] The high-level abstract features are passed to the fully connected layer, and the weight matrix is ​​used to map the high-level abstract features to different impurity types and then the Softmax function is used to transform them into probability distributions.

[0029] Output the specific impurity type label based on the category corresponding to the highest probability.

[0030] As a preferred embodiment of the intelligent impurity removal control method of the present invention, the method includes the following steps: selecting the most suitable cleaning mode from a multi-mode cleaning mode library according to the type of impurity.

[0031] A multi-mode cleaning library is constructed based on specific impurity type labels;

[0032] Match the most suitable cleaning mode according to the different types of impurities.

[0033] As a preferred embodiment of the intelligent impurity removal control method described in this invention, the method includes the following steps: using a genetic algorithm to optimize the specific configuration of cleaning parameters in the cleaning mode, transmitting the optimized cleaning parameters to the actuator to start the cleaning program, continuously monitoring the operating status during the cleaning process, and obtaining a cleaning effect evaluation report.

[0034] Initialize the cleaning parameters of the cleaning mode and combine them into a cleaning parameter set;

[0035] Based on the set of cleaning parameters, a genetic algorithm is used to generate M individuals as the initial population.

[0036] The impurity removal efficiency, equipment utilization rate, and environmental impact of the impurity removal equipment were statistically analyzed. A fitness function was defined and the fitness value of each individual was calculated.

[0037] Based on the fitness value of each individual, a selection probability is assigned to each individual using the roulette wheel selection method, and a fitness threshold is set. Individuals that exceed the fitness threshold are regarded as individuals with higher fitness and are randomly sampled into the next generation of the population.

[0038] In the next generation of the population, individuals with high fitness are paired up and a crossover point is randomly selected. After the crossover point is swapped, some parameters are mutated simultaneously until each individual reaches convergence, and the individual with the highest fitness value is obtained.

[0039] The individual with the highest fitness value is used as the optimized cleaning parameter;

[0040] The optimized cleaning parameters are transmitted to the actuator and the corresponding cleaning mode is started for cleaning. During the cleaning process, the core indicators of the cleaning equipment are continuously monitored and fine-tuned according to the actual situation.

[0041] After cleaning is completed, cleaning samples are collected and the cleaning effect is evaluated. The cleaning effect is then compiled into a cleaning effect evaluation report.

[0042] Secondly, the present invention provides an intelligent impurity removal control system, including a data acquisition module for acquiring operational data and preprocessing it to obtain a standardized dataset;

[0043] The parameter output module uses time series analysis and deep learning models to mine the parameter variation patterns in the standardized dataset, and combines the Q-learning algorithm to build an adaptive learning model to output the optimal parameter combination;

[0044] The cleaning selection module uses multimodal sensors to detect impurity characteristics based on the optimal parameter combination, and uses a deep learning model to classify and identify the impurity characteristics, generating specific impurity type labels. Based on the impurity type, it selects the most suitable cleaning mode from a multi-mode cleaning mode library.

[0045] The report generation module uses a genetic algorithm to optimize the specific configuration of cleaning parameters in the cleaning mode, transmits the optimized cleaning parameters to the actuator to start the cleaning program, continuously monitors the operating status during the cleaning process, and obtains a cleaning effect evaluation report.

[0046] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the intelligent noise removal control method as described in the first aspect of the present invention.

[0047] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent noise removal control method as described in the first aspect of the present invention.

[0048] The beneficial effects of this invention are as follows: It uses a genetic algorithm to optimize the specific configuration of cleaning parameters in the cleaning mode, transmits the optimized cleaning parameters to the actuator to start the cleaning program, continuously monitors the operating status during the cleaning process, and obtains a cleaning effect evaluation report. By defining a multi-objective fitness function that considers impurity removal efficiency, equipment utilization rate, and environmental impact, the global search capability of the genetic algorithm resolves the multi-objective conflicts of the cleaning parameters. While ensuring the impurity removal effect, it achieves a comprehensive optimization of reducing reagent consumption and reducing equipment wear, while maintaining the stability of the cleaning process through a dynamic fine-tuning mechanism, thus promoting the development of cleaning and impurity removal technology towards a green and efficient direction. Attached Figure Description

[0049] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of the multi-stage closed-loop control in Example 1;

[0051] Figure 2 The flowchart of the multi-objective parameter optimization genetic algorithm in Example 1 is shown below;

[0052] Figure 3 This is a flowchart of the impurity classification process using multimodal feature fusion in Example 1. Detailed Implementation

[0053] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0055] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0056] Example 1, referring to Figures 1-3 This is the first embodiment of the present invention, which provides an intelligent impurity removal control method, including the following steps:

[0057] S1. Operating data includes fluid pressure, water flow velocity, water flow temperature, and the pressure difference across the membrane (i.e., the difference in fluid pressure across the water filtration membrane).

[0058] Preprocessing includes detecting and removing outliers;

[0059] Calculate the mean and standard deviation of the running data, and use the 3σ rule to consider any data point that deviates from the mean by more than 3 times the standard deviation as an outlier and delete it;

[0060] The preprocessed running data is standardized (using Z-score standardization to subtract the mean of the corresponding running data from each running data point, and then divide by the standard deviation of the running data) to generate a standardized dataset.

[0061] S2. Use time series analysis and deep learning models to discover the patterns of parameter changes in standardized datasets.

[0062] Includes the following steps,

[0063] By selecting fluid pressure, water flow velocity, and pressure difference across the membrane as input variables, and setting the index of the timestamp (converting the input variables to datetime format), a time series feature data matrix is ​​constructed.

[0064] Choose the time window length (e.g., 5 minutes), segment the input variables according to the time window and combine them into feature vectors, then create an empty two-dimensional matrix (the number of rows equals the number of time windows, and the number of columns equals the number of features in each time window), and fill the feature vectors into the corresponding rows of the two-dimensional matrix. For example, the feature vectors of the first time window are filled into the first row, and so on.

[0065] Input the time series feature data matrix into the LSTM model to generate predicted values ​​for future operating states;

[0066] The time series feature data matrix is ​​divided into a training set (80%) and a test set (20%). Mean squared error is selected as the loss function and Adam as the optimizer. The training set is divided into several small batches (e.g., 32, 64, or 128). Only one small batch of data is used for each update. Forward propagation is performed on the updated batches to obtain the predicted values ​​of the running state and compare them with the time series feature data matrix to obtain the specific value of the loss function. Based on the specific value of the loss function, the gradient of the weights and biases of each layer is calculated layer by layer from the output layer of the LSTM using the chain rule. The Adam optimizer is used to update the weights and biases of the LSTM based on the calculated gradients. The steps of loss function, backpropagation, and gradient update are repeated until all batches are processed. The performance of the LSTM is evaluated by calculating the accuracy, recall, and F1 score of the LSTM using the test set. When the performance on the test set no longer improves, the trained LSTM model is obtained and the predicted values ​​of the future running state are generated based on the memory patterns learned during training.

[0067] By comparing the predicted values ​​of future operating states with the time series feature data matrix (for example, if the predicted value of future operating states is 1.25 bar, while the actual value of the time series feature data matrix is ​​1.30 bar, the difference is -0.05 bar), the trend of parameter changes in the standardized dataset can be identified.

[0068] The trend of parameter change over time is evaluated by fitting a straight line. If the slope is positive, it indicates that the parameter is on an upward trend. If the slope is negative, it indicates that the parameter is on a downward trend. (For example, if the predicted value of future operating status shows that the pressure difference across the membrane continues to rise, and the time series feature data matrix also shows a similar upward trend, it can be inferred that there is a long-term growth trend.)

[0069] S3. Combine the Q-learning algorithm to construct an adaptive learning model and output the optimal parameter combination.

[0070] Includes the following steps,

[0071] The Q-learning algorithm is used to define the state space, action space, and reward function, and to create a Q-table (i.e., a two-dimensional table where rows represent states, columns represent actions, and each cell stores the Q-value of the corresponding state-action pair, with the Q-value set to zero or a random decimal).

[0072] State space: Fluid pressure, water flow velocity, and pressure difference across the membrane are used variables. The three used variables are discretized into a finite number of states (for example, water flow velocity is divided into three levels: slow, medium, and fast). Finally, the discretized states of the three used variables are combined to form the state space (for example, if there are 3 levels of fluid pressure and 3 levels of water flow velocity, there are a total of 9 possible states).

[0073] Action space: Determine the executable actions (adjust fluid pressure, change water flow speed, start cleaning program). Similarly, discretize the three actions into a finite number of actions (for example, divide the fluid pressure adjustment action into increase, decrease or keep it unchanged). Combine all actions to form the action space.

[0074] The predicted value of the future running state is input into the state space, and the optimal action is selected from the action space of the Q table based on the current actual running state.

[0075] The ε-greedy strategy sets a positive number ε of 1 to represent the probability of selecting a random action (e.g., 0.1). The remaining 1-ε probability is used to select the currently known best action. A random number between 0 and 1 is generated. If the random number is less than ε, an action is randomly selected. Otherwise, the action with the maximum Q value is selected based on the current actual running state (if multiple actions have the same maximum Q value, one can be randomly selected from them).

[0076] Based on the optimal action and the reward function, the adjusted running state and reward value are recorded, the Q value in the Q table is updated, and the loop is iterated until the Q table converges, thus obtaining the optimal parameter combination.

[0077] Execute the selected optimal action (assuming the selected action is to increase fluid pressure, then adjust the fluid pressure setting accordingly), and record the adjusted operating status, such as (high fluid pressure, low water flow velocity, small pressure difference across the membrane) and the immediate reward value obtained after executing the action, such as an immediate reward of +10 points.

[0078] After execution, the corresponding Q value is found in the Q table. For example, assuming the current state is (medium fluid pressure, medium water flow velocity, pressure difference across the small membrane), and the action is to increase the fluid pressure, the corresponding state-action pair Q value is found in the Q table, and the current state is then updated.

[0079] Based on the new state, find the maximum Q value of the next state in the Q table, and update the Q value of the current state-action pair by combining the learning rate (i.e., the weight that controls the new and old information) and the discount factor (i.e. the importance of future rewards). (Assuming the learning rate is 0.1 and the discount factor is 0.9, the new Q value will be adjusted to a value between the original Q value and the sum of the immediate reward and the future reward).

[0080] S4. Based on the optimal parameter combination, use a multimodal sensor to detect impurity characteristics, and use a deep learning model to classify and identify the impurity characteristics, generating specific impurity type labels.

[0081] Includes the following steps,

[0082] Multimodal sensors include particle size analyzers, ultrasonic sensors, pH sensors, and ORP sensors;

[0083] Using a linear particle size analyzer combined with the principle of laser scattering, the intensity distribution of laser light scattered by particles in the water flow is recorded, and the size range and distribution of particles are statistically analyzed to generate a particle size distribution map.

[0084] Prepare the water sample to be tested and introduce it into the sample cell of the linear particle size analyzer. Start the linear particle size analyzer so that the laser beam passes through the water sample in the sample cell. The particles in the water sample will scatter the laser and generate scattered light signals. The detector of the linear particle size analyzer will automatically detect and record the intensity of the scattered light at different angles.

[0085] The particle size is inferred using Mie scattering theory (applicable to spherical particles) or Rayleigh scattering theory (applicable to small particles). (Larger particles produce stronger scattered light at smaller angles, while smaller particles produce stronger scattered light at larger angles). Based on the inference results, particles are classified according to size. For example, particles can be divided into several size ranges (e.g., 0.1-1 micrometers, 1-10 micrometers, 10-100 micrometers, etc.). For each size range, the number of particles falling within each range is counted (i.e., the frequency distribution). Finally, Matplotlib is used to plot the frequency distribution and size ranges into a bar chart, resulting in a particle size distribution map.

[0086] The ultrasonic sensor is used to emit high-frequency sound waves to the surface of the water filtration membrane and receive the reflected echo signal to calculate the thickness of the deposits on the surface of the water filtration membrane and generate a deposit thickness distribution map.

[0087] When the reflected echo signal is received, the time from the emitted sound wave to each reflection point is recorded (the time at the first reflection point (the filter membrane-attachment interface) is a, and the time at the second reflection point (attachment-air interface) is b). The sound speed (i.e., the known sound speed in water) is combined with the time difference to determine the propagation speed of the sound wave between the surface of the water filter membrane and the attachment layer. Half of b is subtracted from half of a to estimate the thickness of the attachment on the surface of the water filter membrane and multiplied by the propagation speed of the sound wave between the surface of the water filter membrane and the attachment layer to obtain the actual thickness of the attachment on the surface of the water filter membrane.

[0088] pH and ORP sensors are used to monitor changes in pH and ORP values ​​in water bodies to determine whether the water contains organic matter and biofilm, and a water quality chemical characteristic report is generated based on the results.

[0089] When the pH value is maintained within the range of 6.5 to 8.5, it indicates that the acidity or alkalinity of the water body is relatively stable (some scenarios have more specific pH value requirements, such as the pH value of swimming pools, which is usually maintained between 7.2 and 7.8). When the pH value fluctuates significantly, it indicates that there are certain chemical reactions or pollutants in the water body.

[0090] A higher ORP value (200~400mV) usually indicates that the water body has a strong oxidation capacity and a low amount of organic matter and microorganisms, while a lower ORP value (below 200mV) indicates that there is a lot of organic matter or biofilm in the water body, which will consume oxygen and reduce the oxidation capacity of the water body.

[0091] The particle size distribution map, the attachment thickness distribution map, and the water quality chemical characteristics report are combined into a multi-dimensional impurity feature set and converted into image form for input into the CNN model.

[0092] This step combines multiple data sources and uses deep learning models for analysis, which can provide higher detection accuracy and more accurately identify impurities and pollutants in water compared to relying on only one type of sensor data or simple statistical analysis methods.

[0093] CNN models generate feature maps by sliding convolutional kernels across a multidimensional set of impurity features.

[0094] Set the stride of the convolution kernel and place it at the top left corner of the multidimensional impurity feature set, covering a small region (the same size as the kernel). For each value within the small region covered by the kernel (i.e., particle size or attachment thickness at a specific location in the multidimensional impurity feature set), find the corresponding kernel value and pair it. Then multiply this value by the weight at the corresponding position in the kernel (assuming the kernel covers a 3x3 region with 9 values: 1, 2, 3, 4, 5, 6, 7, 8, 9; and the kernel also has 9 weights: 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8). 0.9, then multiply the values ​​in each small region covered by the convolution kernel and the weights of the convolution kernel one by one, and then add them together. According to the set moving step, move the convolution kernel to the right or down one small grid and repeat the weighted summation process until the convolution kernel has traversed the entire multidimensional impurity feature set. Then each convolution kernel will generate a new value and combine them into a feature map.

[0095] The ReLU activation function is used to remove negative values ​​from the feature map and retain positive values. At the same time, max pooling is performed to reduce the resolution of the feature map and obtain high-level abstract features.

[0096] Choose the size of the pooling window (2x2), start from the top left corner of the feature map, cover a small area of ​​the feature map with the pooling window, and then gradually move the window to the right and down until it covers the entire feature map. Find the maximum value of the covered features and record it to form a new and smaller feature map (this feature map has a lower spatial resolution but retains important feature information).

[0097] Repeatedly performing max pooling operations makes the feature map more abstract, forming higher-level abstract features (low-level features (such as edges and textures) are transformed into higher-level features (such as shapes and object parts)).

[0098] The high-level abstract features are passed to the fully connected layer, and the weight matrix is ​​used to map the high-level abstract features to different impurity types and then the Softmax function is used to transform them into probability distributions.

[0099] Assuming there are three impurity types, impurity 1, impurity 2, and impurity 3, the three impurities are combined into an impurity vector. The impurity vector is then multiplied by the weight matrix in the fully connected layer to obtain the matching score for each impurity type. Subsequently, the Softmax function is used to normalize the matching scores for the three impurity types, and the exponent value of each matching score is calculated and summed. Finally, the exponent value of each score is divided by the sum to obtain the final probability distribution (the probability of impurity 1 is exp(S1)÷(exp(S1)+exp(S2)+exp(S3))=0.8, the probability of impurity 2 is exp(S2)÷(exp(S1)+exp(S2)+exp(S3))=0.15, and the probability of impurity 3 is exp(S3)÷(exp(S1)+exp(S2)+exp(S3))=0.05, where S1 represents the probability distribution of impurity 1, S2 represents the probability distribution of impurity 2, and S3 represents the probability distribution of impurity 3).

[0100] Output the specific impurity type label based on the category corresponding to the highest probability;

[0101] The specific impurity types are labeled as suspended solids, biofilms, and organic matter;

[0102] The impurity type labels obtained in this step do not require manual intervention to interpret the probability distribution of the model output. Instead, the system automatically selects the category with the highest probability and outputs the corresponding label, which greatly reduces human error and workload.

[0103] S5. Based on the type of impurities, select the most suitable cleaning mode from the multi-mode cleaning library.

[0104] Includes the following steps,

[0105] A multi-mode cleaning library is constructed based on specific impurity type labels;

[0106] The multi-mode cleaning library includes backwashing mode (using high-pressure water flow to backwash the water filtration membrane to remove particulate matter deposited on the membrane surface), chemical cleaning mode (using chemical reagents such as acids, alkalis, and oxidants to dissolve or remove organic matter or biofilm), and ultrasonic cleaning mode (using the cavitation effect generated by high-frequency vibration to destroy inorganic salts, colloidal particles, oil and grease and other impurities attached to the membrane surface).

[0107] Match the appropriate cleaning mode according to the different types of impurities (if there are multiple possible matches, the most suitable one can be selected according to the specific situation).

[0108] For example, for biofilm impurities, either chemical cleaning or ultrasonic cleaning can be selected. If the biofilm is stubborn, ultrasonic cleaning is preferred; if the biofilm is relatively light, chemical cleaning can be chosen.

[0109] S6. Use a genetic algorithm to optimize the specific configuration of cleaning parameters in the cleaning mode, pass the optimized cleaning parameters to the actuator to start the cleaning program, continuously monitor the running status during the cleaning process, and obtain a cleaning effect evaluation report.

[0110] Includes the following steps,

[0111] Initialize the cleaning parameters of the cleaning mode and combine them into a cleaning parameter set;

[0112] The cleaning parameters for backwashing mode are cleaning time (30 minutes), cleaning frequency (24 hours), and water pressure (0.5 MPa). The cleaning parameters for chemical cleaning mode are cleaning agent type (sodium hypochlorite), agent concentration (1%), chemical reaction time (40~60 minutes), and temperature (around 30℃). The cleaning parameters for ultrasonic cleaning mode are ultrasonic frequency (40kHz), action time (35~45 minutes), and temperature (22~25℃).

[0113] Based on the set of cleaning parameters, a genetic algorithm is used to generate M individuals as the initial population.

[0114] The value of M here is determined based on the specific problem requirements and the required population size, and the common range is between 10 and 100.

[0115] The impurity removal efficiency, equipment utilization rate, and environmental impact of the impurity removal equipment were statistically analyzed. A fitness function was defined, and the fitness value of each individual was calculated (different weights were assigned to the three parameters of impurity removal efficiency, equipment utilization rate, and environmental impact of the impurity removal equipment according to actual needs, and the weights were multiplied and added to obtain the fitness value of each individual).

[0116] Collect data on the amount of impurities that need to be removed from the impurity removal equipment, cleaning time, actual working time, total available time, wastewater treatment volume, and chemical reagent usage.

[0117] The impurity removal efficiency of the equipment is obtained by dividing the amount of impurities to be removed by the total amount of impurities and multiplying by the cleaning time; the equipment utilization rate is obtained by dividing the actual working time of the equipment by the total available time; and the environmental impact on the equipment is obtained by dividing the wastewater treatment volume plus the amount of chemical reagents used by the amount of impurities to be removed.

[0118] Based on the fitness value of each individual, a selection probability is assigned to each individual using the roulette wheel selection method (assuming there are five individuals, namely individual 1, individual 2, individual 3, individual 4, and individual 5, with fitness values ​​of 67.73, 55.40, 89.21, 72.15, and 60.32 respectively, the total fitness value is 344.81, and the selection probability of each individual is obtained by dividing the fitness value of each individual by the total fitness value, such as the selection probability of individual 1 being 0.1964). A fitness threshold is set, and individuals exceeding the fitness threshold are regarded as individuals with higher fitness and enter the next generation of the population by random sampling.

[0119] The range of the fitness threshold needs to be set according to the problem requirements. Choose to set a percentile, that is, the top 20% or top 30% of individuals are considered to be individuals with high fitness.

[0120] The fitness values ​​of the five individuals are sorted from largest to smallest as follows: individual 3 (89.21), individual 4 (72.15), individual 1 (67.73), individual 5 (60.32), and individual 2 (55.40). If the fitness threshold is the top 30% of individuals, then individuals 3 and 4 are considered to have high fitness values.

[0121] In the next generation of the population, individuals with high fitness are paired (two individuals form a pair, called the parent individuals), and a crossover point is randomly selected. After the crossover point, some parameters are exchanged and mutation is performed simultaneously until each individual reaches convergence, at which point the individual with the highest fitness value is obtained.

[0122] The cleaning parameters for each individual are cleaning time, cleaning frequency, cleaning agent type, cleaning agent concentration, water flow pressure, ultrasonic frequency, and temperature. The third parameter (i.e., cleaning agent type) is randomly selected as the crossover point. The parameters of the two parent individuals after the crossover point are exchanged to generate two new offspring individuals (for example, parent individual A's parameters are [45 minutes, every 48 hours, sodium hypochlorite, 2%, 5 m / s, 35 kHz, 28°C], and parent individual B's parameters are [50 minutes, every 24 hours, sulfuric acid, 1.5%, 6 m / s, 40 kHz, 30°C]. If the cleaning agent type is selected as the crossover point, the new offspring individuals after the exchange are offspring individual 1 [45 minutes, every 48 hours, sulfuric acid, 1.5%, 6 m / s, 40 kHz, 30°C] and offspring individual 2 [50 minutes, every 24 hours, sodium hypochlorite, 1.2%, 5 m / s, 35 kHz, 28°C]).

[0123] Set a small mutation probability (e.g., 0.01). For each newly generated offspring individual, randomly select one or more parameters as mutation points (e.g., select the washout time), and adjust the parameter values ​​corresponding to the mutation points (e.g., assuming the washout time for offspring individual 1 is 45 minutes, the mutation operation can be adjusted to 46 minutes or 44 minutes).

[0124] The individual with the highest fitness value was used as the optimized cleaning parameters (among the five individuals listed, individual 3 had the highest fitness value of 89.21, and individual 3 was selected as the optimal cleaning parameters [60 minutes, once every 72 hours, sodium hypochlorite, 1%, 4 m / s, 30 kHz, 25°C]).

[0125] The optimized cleaning parameters are transmitted to the actuator (i.e., the cleaning equipment control system) and the corresponding cleaning mode is started for cleaning. During the cleaning process, the core indicators of the cleaning equipment are continuously monitored and fine-tuned according to the actual situation.

[0126] The core performance indicators for cleaning equipment during operation include impurity removal efficiency and water quality parameters (pH value, conductivity, turbidity).

[0127] When the impurity removal efficiency is low, gradually increase the cleaning agent concentration (1.05% or 1.1%) and extend the cleaning time (extend to 63 minutes, i.e., increase by 5% or 66 minutes, i.e., increase by 10%) according to the actual situation. When the water quality parameters exceed the standard range (pH value is 6.5~8.5, conductivity is 200μS / cm, turbidity is less than 1NTU), if the pH value is less than 6.5, add sodium hydroxide to increase the pH value; if it is greater than 8.5, add sulfuric acid to decrease the pH value; when the conductivity exceeds 200μS / cm, add sodium chloride solution to increase the conductivity to the required range; when the turbidity exceeds 1NTU, a more efficient filter element needs to be replaced and the number of filtrations increased to reduce the turbidity to below 1NTU.

[0128] After cleaning is completed, cleaning samples are collected and the cleaning effect is evaluated (such as measuring the impurity content and water quality parameters before and after cleaning; assuming that the turbidity of the water sample before cleaning is 5 NTU and the turbidity of the water sample after cleaning is 1 NTU, the proportion of impurities removed can be calculated as 5-1 divided by 5, which is 80%). The cleaning effect is then integrated into a cleaning effect evaluation report.

[0129] The cleaning effect evaluation report includes cleaning parameters (such as water flow pressure, cleaning agent concentration, etc.), changes in water quality parameters, impurity removal ratio, and fine-tuning results;

[0130] Compared to existing impurity removal technologies that only focus on a single indicator (such as the impurity removal rate), this cleaning effect evaluation report covers multiple aspects such as the impurity removal rate and water quality parameters, providing a more comprehensive and detailed assessment. This helps managers make more scientific and reasonable decisions, optimize cleaning plans, and thus significantly improve cleaning results.

[0131] This embodiment also provides an intelligent impurity removal control system, including: a data acquisition module, which collects operating data and preprocesses it to obtain a standardized dataset;

[0132] The parameter output module uses time series analysis and deep learning models to mine the parameter variation patterns in the standardized dataset, and combines the Q-learning algorithm to build an adaptive learning model to output the optimal parameter combination;

[0133] The cleaning selection module uses multimodal sensors to detect impurity characteristics based on the optimal parameter combination, and uses a deep learning model to classify and identify the impurity characteristics, generating specific impurity type labels. Based on the impurity type, it selects the most suitable cleaning mode from a multi-mode cleaning mode library.

[0134] The report generation module uses a genetic algorithm to optimize the specific configuration of cleaning parameters in the cleaning mode, transmits the optimized cleaning parameters to the actuator to start the cleaning program, continuously monitors the operating status during the cleaning process, and obtains a cleaning effect evaluation report.

[0135] This embodiment also provides a computer device applicable to the intelligent impurity removal control method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the intelligent impurity removal control method proposed in the above embodiment.

[0136] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0137] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the intelligent noise reduction control method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0138] In summary, this invention optimizes the specific configuration of cleaning parameters in a cleaning mode using a genetic algorithm, transmits the optimized parameters to the actuator to initiate the cleaning program, continuously monitors the operating status during the cleaning process, and obtains a cleaning effect evaluation report. By defining a multi-objective fitness function encompassing impurity removal efficiency, equipment utilization, and environmental impact, the global search capability of the genetic algorithm resolves the multi-objective conflicts of the cleaning parameters. While ensuring impurity removal effectiveness, it achieves a comprehensive optimization that reduces reagent consumption and equipment wear, while maintaining the stability of the cleaning process through a dynamic fine-tuning mechanism, thus promoting the development of cleaning and impurity removal technology towards a green and efficient direction.

[0139] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A control method of intelligent impurity removal, characterized by: The application relates to a method for optimizing cleaning parameters of a multi-element cleaning mode. Collecting operation data and performing preprocessing to obtain a standardized data set; Using a time series analysis method and a deep learning model to mine parameter variation rules in the standardized data set, and combining a Q-learning algorithm to construct an adaptive learning model to output an optimal parameter combination; According to the optimal parameter combination, using a multi-modal sensor to detect impurity characteristics, and using a deep learning model to classify and identify the impurity characteristics to generate specific impurity type labels, and selecting the most suitable cleaning mode from a multi-element cleaning mode library according to the impurity type; Using a genetic algorithm to optimize the specific configuration of the cleaning parameters in the cleaning mode, and transmitting the optimized cleaning parameters to an executing mechanism to start a cleaning program, and continuously monitoring the operation state during the cleaning process to obtain a cleaning effect evaluation report; The method for optimizing the specific configuration of the cleaning parameters in the cleaning mode comprises the following steps: Initializing the cleaning parameters of the cleaning mode and combining the cleaning parameters into a cleaning parameter set; Based on the cleaning parameter set, using a genetic algorithm to generate M individuals as an initial population; Statistically obtaining the impurity removal efficiency of the impurity removal equipment, the equipment utilization rate and the influence degree of the environment on the impurity removal equipment, defining a fitness function and calculating the fitness value of each individual, and according to actual requirements, assigning different weights to the impurity removal efficiency, the equipment utilization rate and the influence degree of the environment on the impurity removal equipment and multiplying and adding them to obtain the fitness value of each individual; According to the fitness value of each individual, using a roulette selection method to assign a selection probability to each individual, and setting a fitness threshold, regarding the individuals exceeding the fitness threshold as individuals with higher fitness and entering the next generation population in a random sampling manner; In the next generation population, the individuals with higher fitness are paired and a crossover point is randomly selected, the parameters after the crossover point are exchanged and mutation operation is performed, until each individual converges, and the individual with the highest fitness value is obtained; The individual with the highest fitness value is used as the optimized cleaning parameter.

2. The control method of the intelligent impurity removal according to claim 1, characterized in that: The preprocessing comprises detecting abnormal values and removing the abnormal values; The preprocessed operation data is standardized to generate a standardized data set.

3. The control method of claim 2, wherein: Using a time series analysis method and a deep learning model to mine parameter variation rules in the standardized data set comprises the following steps: Selecting fluid pressure, water flow velocity and pressure difference on both sides of the membrane as input variables, and setting an index of time stamp to construct a time series feature data matrix; Inputting the time series feature data matrix into an LSTM model to generate a prediction value of future operation state; Comparing the prediction value of the future operation state with the time series feature data matrix to identify the trend rule of parameter variation in the standardized data set.

4. The control method of claim 3, wherein: Combining a Q-learning algorithm to construct an adaptive learning model to output an optimal parameter combination comprises the following steps: Using a Q-learning algorithm to define a state space, an action space and a reward function, and creating a Q table; Inputting the prediction value of the future operation state into the state space, and using a greedy strategy to select an optimal action from the action space of the Q table according to the current actual operation state; According to the optimal action combined with the reward function, the adjusted running state and reward value are recorded, the Q value in the Q table is updated, and the loop iteration is performed until the Q table converges, and the optimal parameter combination is obtained.

5. The control method of claim 4, wherein: According to the optimal parameter combination, the impurity characteristics are detected by using the multi-modal sensor, and the deep learning model is used to classify and identify the impurity characteristics to generate specific impurity type labels, including the following steps, Using a particle size analyzer combined with laser scattering principle, record the scattering light intensity distribution of particles in water flow to the laser and count the size range and distribution of particles, generate particle size distribution graph; Use the ultrasonic sensor to emit high-frequency sound waves to the surface of the water quality filter membrane, and receive the reflected echo signal, calculate the thickness of the attached matter on the surface of the water quality filter membrane, and generate an attached matter thickness distribution graph; Use pH sensor and ORP sensor to monitor the change of pH value and ORP value of water body, judge whether there is organic matter and biofilm in water body, and generate water quality chemical property report according to the judgment result; Combine the particle size distribution graph, attached matter thickness distribution graph and water quality chemical property report into a multi-dimensional impurity feature set and convert it into an image form input into the CNN model; The CNN model slides and scans on the multi-dimensional impurity feature set through the convolution kernel to generate a feature map; Using ReLU activation function to remove negative values and retain positive values of the feature map, while performing max pooling operation to reduce the resolution of the feature map, and obtaining high-level abstract features; The high-level abstract features are transmitted to the fully connected layer, and the weight matrix is used to map the high-level abstract features to different impurity types and use the Softmax function to convert them into probability distribution; According to the category corresponding to the highest probability, output the specific impurity type label.

6. The control method of the intelligent impurity removal according to claim 5, characterized in that: According to the impurity type, select the most suitable cleaning mode from the multi-element cleaning mode library, including the following steps, According to the specific impurity type label, construct a multi-element cleaning mode library; According to the different types of impurities, match the corresponding cleaning mode.

7. The control method of claim 6, wherein: Use genetic algorithm to optimize the specific configuration of cleaning parameters in the cleaning mode, pass the optimized cleaning parameters to the actuator to start the cleaning program, and continuously monitor the running state during the cleaning process to obtain a cleaning effect evaluation report, including the following steps, Initialize the cleaning parameters of the cleaning mode and combine them into a cleaning parameter set; Based on the cleaning parameter set, use genetic algorithm to generate M individuals as the initial population; Statistical the impurity removal efficiency of the impurity removal equipment, the equipment utilization rate and the influence degree of the environment on the impurity removal equipment, define the fitness function and calculate the fitness value of each individual; According to the fitness value of each individual, use roulette selection method to assign a selection probability to each individual, and set a fitness threshold, individuals exceeding the fitness threshold are considered to have higher fitness and enter the next generation population in a random sampling manner; In the next generation population, the individuals with higher fitness are paired and a crossover point is randomly selected, the parameters after the crossover point are exchanged and mutation operation is performed, until each individual converges, and the individual with the highest fitness value is obtained; The individual with the highest fitness value is used as the optimized cleaning parameter; The optimized cleaning parameters are transmitted to the actuators and the corresponding cleaning mode is started for cleaning. During the cleaning process, the core indicators during the operation of the cleaning equipment are continuously monitored and fine-tuned according to the actual situation. After cleaning, the cleaning samples are collected and the cleaning effect is evaluated, and the cleaning effect is integrated into a cleaning effect evaluation report.

8. A control system for intelligent impurity removal based on the control method for intelligent impurity removal according to any one of claims 1 to 7, characterized in that: It comprises, a data acquisition module that acquires operation data and pre-processes the operation data to obtain a standardized data set; a parameter output module that uses time series analysis and deep learning models to mine parameter variation rules in the standardized data set, and uses a Q-learning algorithm to construct an adaptive learning model to output an optimal parameter combination; a cleaning selection module that uses multi-modal sensors to detect impurity characteristics according to the optimal parameter combination, and uses a deep learning model to classify and identify the impurity characteristics to generate specific impurity type labels, and selects the most suitable cleaning mode from a multi-element cleaning mode library according to the impurity type; a report generation module that uses a genetic algorithm to optimize the specific configuration of the cleaning parameters in the cleaning mode, transmits the optimized cleaning parameters to the actuators to start the cleaning program, and continuously monitors the operating state during the cleaning process to obtain a cleaning effect evaluation report. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to implement the steps of the intelligent impurity removal control method of any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to implement the steps of the intelligent impurity removal control method of any one of claims 1-7.

Images