A method and device for detecting adsorption performance of activated carbon based on temperature and humidity compensation

Abstract

The application discloses a kind of based on temperature and humidity compensation activated carbon adsorption performance detection method and device;The activated carbon adsorption performance detection method is by obtaining the key parameter of measured activated carbon in the process of adsorbing waste gas, and based on the in-situ detection of activated carbon adsorption performance of sample data set is constructed to key parameter;At the same time, based on the resistance of measured activated carbon changes with the temperature and humidity of environment to construct temperature and humidity compensation model;The temperature and humidity of environment in sample data set is calculated using temperature and humidity compensation model, and the resistance corresponding to temperature and humidity in each sample is added to sample data set as compensation resistance, improve the accuracy of activated carbon adsorption performance detection;In addition, the random forest model with key parameters and compensation resistance as input, activated carbon adsorption performance as output is constructed, and the random forest model is trained using sample data set;The adsorption performance detection of measured activated carbon is carried out using trained random forest model.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of detection and measurement, and particularly relates to a kind of active carbon adsorption performance detection method and device based on temperature and humidity compensation. BACKGROUND

[0002] With the improvement of material life level, the public is increasingly concerned about the health problems caused by environmental pollution, especially the problem of air pollution. In air pollution, the harm of volatile organic compounds (VOCs) is particularly prominent. Due to its excellent adsorption performance, activated carbon has become an important material for reducing VOCs emissions. Currently, many enterprises use activated carbon to adsorb VOCs in industrial waste gas. However, activated carbon will reach saturation state after adsorbing industrial waste gas, thereby losing adsorption capacity. The current mainstream evaluation methods of activated carbon adsorption performance include iodine value determination method, carbon tetrachloride (CTC) adsorption capacity detection method and benzene adsorption capacity detection method. These methods can provide relatively accurate evaluation of activated carbon adsorption performance. However, these methods require collecting activated carbon samples from activated carbon adsorption boxes and sending them to professional testing agencies or laboratories for detailed analysis. The whole detection process is complicated and time-consuming, usually taking 2 to 3 days, which seriously affects the detection efficiency and cannot achieve the purpose of timely determining the performance of activated carbon. SUMMARY

[0003] The present application provides a kind of active carbon adsorption performance detection method and device based on temperature and humidity compensation.

[0004] In the first aspect, the present application provides a kind of active carbon adsorption performance detection method based on temperature and humidity compensation, which includes:

[0005] Obtaining key parameters of the measured activated carbon during the adsorption of waste gas, and constructing a sample data set based on the key parameters; the key parameters include the resistance value, weight of the measured activated carbon, temperature and humidity of the environment, wind speed and volatile organic compound concentration of the waste gas;

[0006] Based on the change of the resistance value of the measured activated carbon with the temperature and humidity of the environment, a temperature and humidity compensation model is constructed;

[0007] The temperature and humidity compensation model is used to calculate the resistance value corresponding to the temperature and humidity in each sample by using the temperature and humidity of the environment in the sample data set, and the resistance value is added to the sample data set as a compensation resistance value;

[0008] A random forest model is constructed with the key parameters and compensation resistance value as input and the activated carbon adsorption performance as output, and the random forest model is trained using the sample data set; the trained random forest model is used for adsorption performance detection of the measured activated carbon.

[0009] As preferred, the parameters of the random forest model are optimized using the particle swarm algorithm, and the process is as follows:

[0010] The particle swarm is initialized, and when initializing the positions of the particles in the particle swarm, the particles in the particle swarm are divided into two parts, one part is initialized based on the importance of the random forest model parameters, and the other part is randomly initialized within the parameter boundary; the particles in the particle swarm are iterated multiple times, and the speed and position of the particles are updated based on the fitness of the particles in each iteration until the iteration termination condition is met, and the particle with the optimal fitness is taken as the global optimal solution; and based on the global optimal solution obtained by the particle swarm algorithm, the parameter combination of the random forest model is obtained.

[0011] As preferred, in the particle iteration process, the method for updating the particle speed is as follows:

[0012] The speed of the particle in the next iteration is obtained using the speed update formula of the particle swarm based on the fitness of the particle; if the iteration number t is greater than half of the maximum iteration number and meets the multiple of the preset value, the particle speed is further disturbed and updated, and the disturbed and updated speed is taken as the speed of the particle in the next iteration; the disturbed and updated speed is expressed as:

[0013]

[0014]

[0015] wherein, is the speed obtained using the speed update formula; is the disturbance intensity; T is the mapping function; x is a random number; and are the lower limit and the upper limit of the th parameter of the random forest model, respectively. j

[0016] As preferred, a dynamic inertia weight and a learning factor are introduced in the speed update formula of the particle swarm.

[0017] As preferred, after the particle swarm algorithm obtains the global optimal solution, the downhill simplex method is used to locally search the global optimal solution to obtain the local optimal solution corresponding to the global optimal solution, and the local optimal solution is taken as the parameter combination of the random forest model.

[0018] As preferred, the optimized parameters of the random forest model include , , , and the maximum number of features.​

[0019] As preferred, the data of the key parameters is preprocessed before the sample data set is constructed, and the specific process is as follows:

[0020] The data of the key parameters is divided into data of multiple sampling periods; the lower quartile Q1 and the upper quartile Q3 of the data in each sampling period are calculated respectively, the lower bound L and the upper bound U of the outliers are constructed based on the lower quartile Q1 and the upper quartile Q3, the data in the sampling period exceeding the lower bound L and the upper bound U is removed, and the average value of the remaining data in the sampling period is calculated as a sample. L U

[0021] As preferred, after the compensation resistance value is added to the sample data set, the sample data set is augmented, and the augmentation method is as follows:

[0022] A key parameter matrix is constructed with the data in the sample data set, new data is obtained by injecting noise into the key parameter matrix, the new data is added to the sample data set, and the augmentation of the sample data set is completed.

[0023] As preferred, before the sample data set is used to train the random forest model, the label data in the sample data set is converted to approximate normal distribution through Box-Cox transformation.

[0024] In the second aspect, the present application provides an activated carbon adsorption performance detection device based on temperature and humidity compensation, which is used to execute the activated carbon adsorption performance detection method described above; the activated carbon adsorption performance detection device comprises a data acquisition module, a control module and a performance detection module; the data acquisition module comprises a detection box, a state detection module and an environment detection module; the detection box is used to store the measured activated carbon; the state detection module comprises a weight sensor and a resistance detection module; the weight sensor is installed at the bottom of the detection box; the resistance detection module comprises two graphite electrode plates and a resistance transmitter; the two graphite electrode plates are respectively installed on one opposite side of the inside of the detection box, and the measured activated carbon is filled between the two graphite electrode plates; the two detection ends of the resistance transmitter are respectively connected to the side of the two graphite electrode plates close to the measured activated carbon;

[0025] The environment detection module comprises a temperature and humidity sensor, a wind speed sensor and a volatile organic compound detection sensor; the control module is used to transmit the data collected by the data acquisition module to the performance detection module; the performance detection module is used to detect the activated carbon adsorption performance according to the data collected by the data acquisition module.

[0026] The present application has the beneficial effects that:

[0027] ​​1、The present application constructs a random forest model to detect the adsorption performance of activated carbon in situ by collecting key parameters of activated carbon in the process of adsorbing waste gas, solves the problem of complicated and time-consuming activated carbon adsorption performance detection process, improves the convenience of adsorption performance detection; at the same time, the present application introduces a compensation resistance as the input of the random forest, reduces the implicit influence of temperature and humidity change on the resistance, and improves the accuracy of activated carbon adsorption performance detection.

[0028] 2、The present application uses particle swarm optimization algorithm to optimize the parameters in the random forest model, processes the particle position by combining parameter importance and random initialization, so that the particle is more likely to approach the optimal solution space in the initial stage; at the same time, after the global optimal solution is obtained by the particle swarm optimization algorithm, the downhill simplex method is used for local search of the global optimal solution, and the parameters such as tree depth and split sample number which are most sensitive to model complexity are focused on, which can effectively improve the performance of the model. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 It is a schematic diagram of the activated carbon adsorption performance detection device in embodiment 1 of the present application.

[0030] Figure 2 It is a sectional view of the activated carbon filled between the two graphite electrode plates in embodiment 1 of the present application.

[0031] Figure 3 It is a schematic diagram of the activated carbon adsorption performance detection method in embodiment 2 of the present application.

[0032] Figure 4 It is a schematic diagram of the change of iodine value and activated carbon resistance with time in embodiment 2 of the present application.

[0033] Figure 5 It is a schematic diagram of the change of activated carbon resistance with temperature under the same humidity in embodiment 2 of the present application.

[0034] Figure 6 It is a schematic diagram of the change of activated carbon resistance with temperature under different humidity in embodiment 2 of the present application.

[0035] The figure mark: 1, detection box; 2, weight sensor; 3, graphite electrode plate; 4, resistance transmitter; 5, measured activated carbon; 6, temperature and humidity sensor; 7, wind speed sensor; 8, volatile organic compound detection sensor; 9, control cabinet. DETAILED DESCRIPTION

[0036] The present application will be further described below in combination with the drawings.

[0037] Embodiment 1

[0038] As Figure 1 And Figure 2As shown, a kind of active carbon adsorption performance detection device based on temperature and humidity compensation includes data acquisition module, control module and performance detection module.Data acquisition module includes detection box 1, state detection module and environmental detection module;Detection box 1 is used to store the measured active carbon 5;A plurality of through holes are formed on one relative side wall of detection box 1, so that the measured active carbon can be better adsorbed by waste gas;Spring pressure plate is provided on the top of detection box 1, for stabilizing the measured active carbon in detection box 1;State detection module includes weight sensor 2 and resistance detection module;Weight sensor 2 is installed at the bottom of detection box 1, for detecting the weight of the measured active carbon;Resistance detection module includes two graphite electrode plates 3 and one resistance transmitter 4;Two graphite electrode plates 3 are respectively installed on the mutually opposite sides of detection box 1, and the side without through hole, and the measured active carbon is uniformly filled between two graphite electrode plates 3;Resistance transmitter 4 is arranged outside detection box 1, and two detection ends thereof are respectively connected to the side of two graphite electrode plates 3 close to the measured active carbon through wires, for detecting the resistance of the measured active carbon.

[0039] Environmental detection module is arranged outside detection box 1, for detecting the environmental state where the measured active carbon is located.Environmental detection module includes temperature and humidity sensor 6, wind speed sensor 7 and volatile organic compound detection sensor 8.Temperature and humidity sensor 6 is used to detect the temperature and humidity of the environment where the measured active carbon is located;Wind speed sensor 7 is used to detect the wind speed of waste gas treated by the measured active carbon;Volatile organic compound detection sensor 8 is used to detect the content of volatile organic compounds in waste gas treated by the measured active carbon.

[0040] Control module includes control cabinet 9, leakage protection switch, analog module, gateway, signal amplifier and power supply module installed on control cabinet 9;

[0041] Analog module is used to convert analog signal detected by data acquisition module into digital signal;Controller is used to convert digital signal into real-time data, and input into performance detection module through gateway and signal amplifier;Performance detection module is used to obtain iodine value or CTC value of the measured active carbon according to the data collected by data acquisition module, and obtain the adsorption performance of active carbon based on iodine value or CTC value.Leakage protection switch is used to automatically cut off power supply when leakage occurs in circuit;Power supply module is used to supply power for active carbon adsorption performance detection device.

[0042] In this embodiment, data acquisition module is connected with analog module and power supply module through wires, and wires are placed in bellows, which can prevent damage to external wires;Data acquisition module outputs 4-20mA signal;Controller has 12 digital input ports of 24V DC, 8 digital output ports of 24V DC, 6 expandable expansion modules and 1 signal board, and can support maximum 256 point switch quantity I / O and 49 point analog quantity I / O.

[0043] In some embodiments, the resistance transmitter 4 is spaced from the external environment by a protective cover for preventing loss of the device.

[0044] Embodiment 2

[0045] As shown in Figure 3 , a method for detecting the adsorption performance of activated carbon based on temperature and humidity compensation, using the activated carbon adsorption performance detection device in embodiment 1, the method for detecting the adsorption performance of activated carbon includes the following steps:

[0046] Step one, using the activated carbon adsorption performance detection device to collect the key parameters of the measured activated carbon in the process of adsorbing waste gas in real time; the key parameters include the weight, resistance value of the measured activated carbon, the temperature and humidity of the environment, the wind speed and volatile organic compound concentration of the waste gas. In order to obtain accurate data, the collected data is screened, and the specific process is as follows:

[0047] According to the collection time point of the daily activated carbon sample, the data within 10 minutes before and after is screened; the lower quartile Q1 (25th percentile) and the upper quartile Q3 (75th percentile) of the data are calculated, and the interquartile range IQR is obtained by the difference between the two; based on the lower quartile Q1, the upper quartile Q3 and the interquartile range IQR, the lower bound L and the upper bound U of the outlier are constructed, which are represented as follows:

[0048]

[0049]

[0050] The data exceeding the lower bound L and the upper bound U is regarded as an outlier, and the outliers are removed, and the average value of the remaining data is calculated as a sample for constructing the data set. By accumulating the collection times of the activated carbon samples, a sample data set with a certain scale corresponding to the collection times can be obtained; the sample data set is labeled, and the iodine value is used as the evaluation index of the adsorption performance of the measured activated carbon.

[0051] Step two, as shown in Figure 4 , the resistance value of activated carbon has a strong relationship with the iodine value, so it is necessary to explore the related characteristics of the resistance value of activated carbon; since the resistance value of activated carbon is significantly affected by the temperature and humidity of the environment, a constant temperature and humidity chamber is used to establish the relationship between the resistance value of activated carbon and the temperature and humidity of the environment, as shown in Figure 5 and Figure 6 From Figure 5 , it can be seen that there is a significant positive correlation between the temperature of the environment and the resistance value of activated carbon. From Figure 6It can be seen that for the three humidity gradients of 50% RH, 65% RH, and 80% RH, and within each humidity gradient, the temperature increases from 10°C to 30°C in steps of 2°C, and under the same temperature conditions, the corresponding resistance also changes with humidity. At the same time, when the ambient temperature increases in steps of 2°C, the active carbon material in the device needs to go through a relaxation process of about 40 minutes, and the temperature field and resistance value reach a steady state balance synchronously. In addition, at each given time point, the time required for the humidity to reach saturation is constant at 24 hours, and under changing humidity conditions, the active carbon adsorbs water vapor in the air for a long time. During the adsorption process, as water vapor molecules gradually occupy the pores of the active carbon to form new conductive channels, the resistance of the active carbon shows a gradual downward trend.

[0052] Considering the influence of temperature and humidity on resistance value, a temperature and humidity compensation model is established using quadratic polynomial regression technology, i.e., an equation with temperature and humidity as independent variables and resistance value as dependent variable. The resistance value corresponding to the temperature and humidity in each sample is calculated using the temperature and humidity compensation model, and this resistance value is taken as the compensation resistance value. To further enrich the data characteristics and improve the model performance, the ideal resistance value is taken as a new feature dimension and integrated into the sample data set. The additional information contained in the compensation resistance value provides a more comprehensive and representative data basis for subsequent data analysis and model training, thereby enhancing the model's ability to capture complex data patterns and improving prediction accuracy.

[0053] Step three, in order to expand the size of the data set and improve the accuracy of the model, the sample data set needs to be expanded by Gaussian matrix. The specific process is as follows:

[0054] A key parameter matrix is constructed using the data in the sample data set, and noise is injected into the key parameter matrix. The key parameter matrix after injecting noise is represented as:

[0055] = W+P

[0056] wherein, W is the key parameter matrix before injecting noise; P is the noise matrix in the injected variable.

[0057] The key parameter matrix after injecting noise is taken as new data and added to the sample data set, thereby improving the comprehensiveness and richness of the data. The Box-Cox transformation is used to convert the skewed label data in the sample data set to approximately normal distribution, satisfying the assumption of the regression model on residual error.

[0058] Step four, build a random forest model with key parameters and compensation resistance as input, and activated carbon adsorption performance as output; use particle swarm algorithm to optimize the parameters of the random forest model, the specific process is as follows:

[0059] 4-1. Initialize the number of particles in the particle swarm, the maximum number of iterations, and the position and speed of the particles, each particle represents a set of parameters of the random forest model 、 、 、 and the maximum number of features); wherein the method of initializing the particle position is as follows: divide the particles in the particle swarm into two parts, the initial position of a part of the particles is expressed as:

[0060]

[0061] wherein, is the position of the i th particle in the j th dimension; represents the importance score of the j th parameter.

[0062] The position of the other part of the particles is randomly initialized within the parameter boundary to increase the diversity of the population, which is expressed as:

[0063] , ≤ ≤

[0064] wherein, represents a uniform distribution in the interval ; and are the lower limit and upper limit of the j th parameter, respectively.

[0065] By combining the advantages of parameter importance and random initialization, the particles are more likely to approach the optimal solution space in the initial stage.

[0066] 4-2. Train the random forest model based on the current particle corresponding parameter combination, and evaluate its performance as fitness value by leave-one-out cross-validation method;

[0067] 4-3. Update the speed and position according to the fitness value of the particle; since in the process of particle swarm optimization, the traditional speed update formula may fall into local optimum in the later stage of search, in order to enhance the global search ability of the algorithm, a dynamic parameter adjustment mechanism is introduced in the speed update formula, which is expressed as:

[0068] + +

[0069] where, and are the velocity and position of the i th particle at the t th iteration; is the individual historical optimal position of the i th particle at the t th iteration; is the group historical optimal position at the t th iteration; w is the inertia weight; and are learning factors; is an adaptive weight, which is a random number between 0 and 1.

[0070] The inertia weight w and learning factors , are dynamically adjusted with the iteration number t and the maximum iteration number , which are expressed as:

[0071]

[0072]

[0073] This dynamic adjustment mechanism enables the algorithm to have strong global search ability at the early stage of iteration, quickly positioning to the better solution space area; while in the later stage of iteration, with the reduction of inertia weight and the adjustment of learning factor, the algorithm can focus more on local search, fine-tuning the optimal solution, thereby improving the accuracy of the model.

[0074] If the iteration number t is greater than half of the maximum iteration number (t ) and is a multiple of 10, the particle velocity is further disturbed and updated, and the velocity of the updated particle is expressed as:

[0075]

[0076]

[0077] where, is the disturbance intensity; T ( x ) is a mapping function; is a random number between 0 and 1.

[0078] Particle velocity updated by disturbance As the velocity at the next iteration; by updating the particle velocity by disturbance, it can break the stagnation of the algorithm near the local optimal solution, so that the particle can jump out of the local optimum and continue to explore a wider solution space, thereby improving the probability of finding the global optimal solution.

[0079] 4-4. Repeat the above steps until the iteration termination condition is met, and take the particle with the optimal fitness as the global optimal solution; perform local search on the global optimal solution using the Nelder-Mead simplex method (Nelder-Mead algorithm) to obtain the local optimal solution corresponding to the global optimal solution, and train the random forest model using the sample data set based on the parameter combination corresponding to the local optimal solution; the Nelder-Mead simplex method is used for local search, which focuses on refining parameters such as tree depth and split sample number that are most sensitive to model complexity, which can effectively improve the performance of the model.

[0080] Step five, use the trained random forest model to detect the adsorption performance of the measured activated carbon.

[0081] In some embodiments, the CTC value is used as an evaluation index of the adsorption performance of the measured activated carbon.

[0082] Step six, precision verification of the present application is carried out by ablation experiment; 10 independent repeated training is adopted, each run includes data preprocessing, 8:2 training set-test set division and standardization, and non-normal label is processed by Box-Cox transformation. The evaluation indexes include root mean square error (RMSE), mean absolute error (MAE), R² and average training time, and paired t t-test is used to verify the significant difference in performance between the present application and the ordinary particle swarm random forest model, and to reduce the influence of random factors on the results, and the results are shown in Table 1.

[0083] Table 1 Comparison of performance indicators of different methods

[0084]

[0085] As can be seen from Table 1, the present application is significantly better than the control method in prediction accuracy and model generalization ability. Specifically, the RMSE of the present application is 3.834±0.200 MPa, which is 0.7% lower than the ordinary particle swarm model, and the P value of paired t test is 0.0099<0.01, proving that the present application effectively avoids the local optimal problem of the particle swarm algorithm. On R²=0.948 and MAE=2.630 MPa, the present application also performs excellently, indicating that the parameter optimization result can more accurately capture the complex nonlinear relationship of the data. Although the training time=109.3s is slightly high, compared with the default configuration of the random forest model=0.6s, the present application achieves a significant performance improvement of 6.1% in RMSE reduction through reasonable computing overhead, verifying the engineering practicability of the hybrid optimization strategy.

[0086] The above description is merely that of a specific implementation of the present application, which enables those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will accord with the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting the adsorption performance of activated carbon based on temperature and humidity compensation, characterized in that, The application comprises the following steps: Obtaining key parameters of the measured activated carbon during the adsorption of waste gas, and constructing a sample data set based on the key parameters; the key parameters include the resistance value, weight of the measured activated carbon, temperature and humidity of the environment, wind speed and volatile organic compound concentration of the waste gas; Constructing a temperature and humidity compensation model based on the change of the resistance value of the measured activated carbon with the temperature and humidity of the environment; Using the temperature and humidity compensation model to calculate the resistance value corresponding to the temperature and humidity of each sample in the sample data set, and adding the resistance value as a compensation resistance value to the sample data set; Constructing a random forest model with the key parameters and the compensation resistance value as inputs and the adsorption performance of the activated carbon as output, and training the random forest model using the sample data set; and using the trained random forest model to detect the adsorption performance of the measured activated carbon; Optimizing the parameters of the random forest model using a particle swarm algorithm, the process being as follows: Initializing the particle swarm, initializing the position of the particles in the particle swarm, dividing the particles in the particle swarm into two parts, initializing one part based on the importance of the random forest model parameters, and randomly initializing the other part within the parameter boundary; iterating the particles in the particle swarm multiple times, updating the speed and position of the particles based on the fitness of the particles in each iteration until the iteration termination condition is met, and taking the particle with the optimal fitness as the global optimal solution; and obtaining the parameter combination of the random forest model according to the global optimal solution obtained by the particle swarm algorithm; In the particle iteration process, the particle speed is updated as follows: The speed of the particle in the next iteration is obtained based on the fitness of the particle using a speed updating formula of the particle swarm; if the iteration number t is greater than half of the maximum iteration number and meets a preset multiple, the particle speed is further disturbed and updated, to obtain a disturbed and updated speed The speed of the particle in the next iteration is obtained based on the fitness of the particle using a speed updating formula of the particle swarm; if the iteration number t is greater than half of the maximum iteration number and meets a preset multiple, the particle speed is further disturbed and updated, to obtain a disturbed and updated speed is expressed as: ； ； wherein, is the velocity obtained using the velocity update formula; is the disturbance intensity; T(x) is the mapping function; is a random number; and are the lower and upper bounds of the parameters of the jth random forest model, respectively. Introducing a dynamic inertia weight and a learning factor into the speed update formula of the particle swarm; After the particle swarm algorithm obtains the global optimal solution, the Nelder-Mead simplex method is used to locally search the global optimal solution to obtain a local optimal solution corresponding to the global optimal solution, and the local optimal solution is taken as the parameter combination of the random forest model.

2. The method for detecting the adsorption performance of activated carbon based on temperature and humidity compensation according to claim 1, characterized in that: The optimized parameters of the random forest model include the number of trees, the maximum depth, the minimum sample number for splitting, the minimum sample number for leaf nodes, and the maximum number of features.

3. The method for detecting the adsorption performance of activated carbon based on temperature and humidity compensation according to claim 1, characterized in that: Before constructing the sample data set, the data of the key parameters is preprocessed, the specific process being as follows: Divide the data of the key parameters into data of multiple sampling periods; calculate the lower quartile Q1 and the upper quartile Q3 of the data in each sampling period, construct the lower bound L and the upper bound U of the outliers based on the lower quartile Q1 and the upper quartile Q3, remove the data in the sampling period that exceeds the lower bound L and the upper bound U, and calculate the average value of the remaining data in the sampling period as a sample.

4. The method for detecting the adsorption performance of activated carbon based on temperature and humidity compensation according to claim 1, characterized in that: After adding the compensation resistance value to the sample data set, the sample data set is augmented, and the augmentation method is as follows: Construct a key parameter matrix using the data in the sample data set, obtain new data by injecting noise into the key parameter matrix, and add the new data to the sample data set to complete the augmentation of the sample data set.

5. The method for detecting the adsorption performance of activated carbon based on temperature and humidity compensation according to claim 1, characterized in that: Before training the random forest model using the sample data set, the label data in the sample data set is converted to approximate normal distribution through Box-Cox transformation.

6. A device for detecting the adsorption performance of activated carbon based on temperature and humidity compensation, characterized in that: The application discloses a humidity and temperature compensation-based activated carbon adsorption performance detection method and device. The activated carbon adsorption performance detection device comprises a data acquisition module, a control module and a performance detection module. The data acquisition module comprises a detection box (1), a state detection module and an environment detection module. The detection box (1) is used for storing measured activated carbon (5). The state detection module comprises a weight sensor (2) and a resistance detection module. The weight sensor (2) is installed at the bottom of the detection box (1). The resistance detection module comprises two graphite electrode plates (3) and a resistance transducer (4). The two graphite electrode plates (3) are respectively installed on opposite sides in the detection box (1), and the measured activated carbon is filled between the two graphite electrode plates (3). The two detection ends of the resistance transducer (4) are respectively connected with the two graphite electrode plates (3) close to the measured activated carbon. The environment detection module comprises a temperature and humidity sensor (6), a wind speed sensor (7) and a volatile organic compound detection sensor (8). The control module is used for transmitting data collected by the data acquisition module to the performance detection module. The performance detection module is used for performing activated carbon adsorption performance detection according to the data collected by the data acquisition module.

Images