Internet-of-things gas detection and control instrument assembly valve based on integrated multi-sensing technology

Abstract

The invention relates to the technical field of gas detection and control instruments, in particular to an internet-of-things gas detection and control instrument assembly valve based on an integrated multi-sensing technology, which comprises a data acquisition module and a control module, the data processing module comprises a plurality of denoising modules and a plurality of weighted average modules, the same type of sensors are in signal connection with the same denoising module at the same time, and the denoising modules are in signal connection with the weighted average modules one by one; and the data fusion module is in signal connection with the plurality of weighted average modules at the same time. According to the design, the same type of data is measured through multiple sensors, the data quality is further improved through the denoising module, then the weighted average module is used for further changing the weights of different sensors according to the accuracy of the sensor data, and finally the data fusion module is used for judging the alarm level in combination with the multiple data. And the accuracy is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of gas detectors, and more specifically, to a valve of an Internet of Things gas detector component based on an integrated multi-sensing technology. Background Art

[0002] With the rapid development of the Internet of Things technology, the demand for precise monitoring and control of various environmental parameters is increasing day by day; in many fields such as industrial production, environmental monitoring, and smart homes, gas detection and monitoring are crucial; traditional gas detection devices often have a single function, can only detect specific types of gases, and have limited accuracy and reliability.

[0003] At the same time, in a complex environment, the data collected by sensors is often affected by noise interference, which affects the accuracy of the detection results; in addition, different types of sensors have differences in performance and characteristics, and the data of a single sensor is difficult to comprehensively and accurately reflect the actual situation. Summary of the Invention

[0004] The purpose of the present invention is to provide a valve of an Internet of Things gas detector component based on an integrated multi-sensing technology to solve the problems raised in the above background art.

[0005] To solve the above technical problems, the present invention provides a valve of an Internet of Things gas detector component based on an integrated multi-sensing technology, including:

[0006] Data acquisition module: The data acquisition module includes several different types of sensors, and there are multiple sensors of the same type and they are installed at different positions;

[0007] Data processing module: The data processing module includes several denoising modules and several weighted average modules. Sensors of the same type are simultaneously connected to the same denoising module in signal, and the denoising module is connected to the weighted average module in signal one by one;

[0008] Data fusion module: The data fusion module is simultaneously connected to several weighted average modules in signal;

[0009] Data transmission module: The data transmission module is connected to the data fusion module in signal.

[0010] As a further improvement of the present technical solution, the specific steps of the denoising module correcting the measurement data of the data acquisition module based on an improved extended Kalman filter algorithm include:

[0011] S1. Obtain the signal data measured by sensors of the same type;

[0012] S2. Establish a state equation and a measurement equation, which are respectively:

[0013]

[0014] Among them, X k+1 is the system state matrix at time k + 1, and f(X k ) is the system state transition equation at time k, W k is the system noise at time k, Z k+1 is the system measurement matrix at time k + 1, and h(X k+1 ) is the system measurement equation at time k + 1, V k is the measurement noise at time k;

[0015] S3. Optimize the variables in the system noise and measurement noise at the initial moment based on the improved whale algorithm;

[0016] S4. Calculate the prior estimate value of X k ;

[0017] S5. Calculate the prior error covariance matrix;

[0018] S6. Calculate the Kalman gain;

[0019] S7. Calculate the posterior estimate value of X k ;

[0020] S8. Update the error covariance matrix.

[0021] As a further improvement of this technical solution, the specific steps for optimizing the variables in the system noise and measurement noise at the initial moment based on the improved whale algorithm in S3 are as follows:

[0022] S31. Initialize parameters: Set the number of whale populations as N, the maximum number of iterations as T, the spatial dimension as D, and the position vector of the whale represents the variables in the system noise and measurement noise;

[0023] S32. Process boundary conditions, process each individual in the population respectively, calculate the coefficient vectors and generate a uniformly distributed decision random number ρ,

[0024]

[0025] Among them, is the control vector, and it decreases from 2 to 0 with the increase of the number of iterations throughout the iteration cycle, is a random vector on [0, 1];

[0026] S33. Calculate the fitness value: Use the mean square error value between the actual measurement value and the filtered output value as the fitness function;

[0027] S34. Predation to find the optimal solution:

[0028] Randomly search for prey: When ρ < 0.5 and At this time, introduce a feedback mechanism.

[0029]

[0030] Among them, t is the current iteration number, Is the position vector of randomly selected whale individuals in the population at the t-th iteration, f rand Is The corresponding whale fitness value, Is the position vector of the whale individual selected after combining the optimal position in the population at the t-th iteration, α is the adjustment coefficient, β is the diversity coefficient, Is the optimal position vector of the objective function in the current population, f best Is The corresponding whale fitness value, f n Is the fitness value corresponding to the n-th whale in the current population, Is the average fitness value of the current population, Represents the position of the current whale individual at the (t + 1)-th iteration, Is the position vector of the current whale individual at the t-th generation;

[0031] Contract and surround the prey: When ρ < 0.5 and At this time, introduce an adaptive weight,

[0032]

[0033] Among them, γ is the adaptive weight, γ0 is the set initial weight, and τ is the adjustment parameter;

[0034] Spiral update position: When ρ ≥ 0.5,

[0035]

[0036] Among them, s is a constant used to define the logarithmic spiral shape, and l is a uniformly distributed random number between [-1, 1];

[0037] S35. Judge the algorithm end condition, and end when the maximum iteration number is reached or the results are the same after a certain number of iterations.

[0038] As a further improvement of this technical solution, the weighted average module weights the data transmitted by the same type of sensors based on information entropy. The specific steps include:

[0039] A1. Let n r A r The data measured by the sensors are respectively corrected by the denoising module and then transmitted into the weighted average module;

[0040] A2. For each A r sensor, count the probability that its correction value appears within h hours. Let the i-th A r sensor have q different values for its correction value, denoted as v i1 , v i2 , …, v ij , … v iq , and the corresponding frequencies be f i1 , f i2 , …, f ij , …, f iq , and the corresponding probability distribution be p i1 , p i2 , …, p ij , … p iq , then

[0041]

[0042] A3. Calculate the information entropy of each A r sensor according to the probability distribution. Then

[0043]

[0044] where H(i) is the information entropy of the i-th A r sensor;

[0045] A4. Normalize the information entropy to obtain a credibility index. Then

[0046]

[0047] where C(i) is the credibility index of the i-th A r sensor;

[0048] A5. Perform weighted average and determine the weight of each A r sensor according to the credibility index. Then

[0049]

[0050] where w i is the weight of the i-th A r sensor;

[0051] A6. Calculate the weighted sum of the corrected data of n r A r sensors.

[0052] As a further improvement of this technical solution, after the weighted average module calculates the weighted sum of the measurement data of sensors of the same type, the data fusion module fuses the data of different types of sensors based on random forest and outputs an alarm level.

[0053] As a further improvement of this technical solution, the specific steps for the data fusion module to fuse different types of sensor data based on random forest are as follows:

[0054] B1. Use historical data as the data set E, divide the data set into a training set E1 and a test set E2. The measurement data of each type of sensor, after being corrected by the denoising module and calculated by the weighted average module, are used as the attributes of the input data, and the alarm level is used as the output data;

[0055] B2. Construct a random forest. Use the Bootstrap algorithm to draw e samples from the training set with replacement to generate K decision trees. When splitting each node of the decision tree, randomly select m attributes from all M attributes, and select the optimal attribute from the m attributes as the splitting object according to the Gini index, and use the classification and regression tree algorithm to construct the decision tree;

[0056] B3. Use the training set data to train the random forest model;

[0057] B4. Use the test set data to evaluate the trained random forest model, and optimize the random forest model according to the evaluation results.

[0058] As a further improvement of this technical solution, the data transmission module is wirelessly connected to an external server, and the data transmitted by the data transmission module includes various gas concentration thresholds, control instructions, environmental parameters, alarm levels, and various gas concentrations.

[0059] Compared with the prior art, the beneficial effects of the present invention are as follows: In the valve of the IoT gas detection and control instrument component based on the integrated multi-sensing technology, by measuring the same type of data with multiple sensors, the data accuracy is improved. Then, the data quality is further improved by the denoising module. After that, the weighted average module is used to further change the weights of different sensors according to the accuracy of the sensor data. Finally, the data fusion module combines various data to judge the alarm level, with higher accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a schematic structural diagram of the present invention;

[0061] Figure 2 is a schematic connection diagram of the data processing module, the data fusion module, and the data transmission module in the present invention;

[0062] The meanings of the various marks in the figure are as follows: 1. Data acquisition module; 2. Data processing module; 20. Denoising module; 21. Weighted average module; 3. Data fusion module; 4. Data transmission module. DETAILED DESCRIPTION OF THE INVENTION

[0063] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0064] As Figure 1 and Figure 2 shown, this embodiment provides a valve for an Internet of Things gas detection and control instrument component based on integrated multi-sensing technology, including:

[0065] Data acquisition module 1: The data acquisition module 1 includes several different types of sensors. There are multiple sensors of the same type and they are installed at different positions. The types of sensors can include temperature sensors, humidity sensors, combustible gas sensors, toxic gas sensors, etc. In addition, when using the sensors, it is best to be equipped with corresponding amplification and filtering circuits;

[0066] Data processing module 2: The data processing module 2 includes several denoising modules 20 and several weighted average modules 21. Sensors of the same type are simultaneously connected to the same denoising module 20 in signal. The data transmitted by sensors of the same type is input into the denoising module 20 as a matrix. The denoising module 20 is connected to the weighted average module 21 in a one-to-one signal connection. Isolation devices such as optocouplers can be used between the front and rear module connections to ensure that the front and rear modules do not affect each other;

[0067] Data fusion module 3: The data fusion module 3 is simultaneously connected to several weighted average modules 21 in signal. The data fusion module 3 aggregates various data, improves the detection accuracy, and adapts to complex and changeable environments;

[0068] Data transmission module 4: The data transmission module 4 is connected to the data fusion module 3 in signal.

[0069] The Extended Kalman Filter is an extension of the traditional Kalman Filter to handle nonlinear systems. It linearizes the nonlinear system around the current estimated value and estimates the state of the system through two steps: prediction and update. In the prediction stage, the state and error covariance at the next moment are predicted according to the system model; in the update stage, the predicted value is corrected using the actual measurement value to obtain a more accurate state estimate. Sensors are often subject to various noises during the measurement process, such as environmental noise, electronic noise, etc. The Extended Kalman Filter can effectively process these noises, smooth and filter the measurement data, reduce the impact of noise on the measurement results, thereby improving the accuracy of the data, and at the same time correct the nonlinear error: for some nonlinear sensor systems, the measurement data may have nonlinear errors. The Extended Kalman Filter can linearize the nonlinear system, thus better estimating the system state, correcting the nonlinear error, and improving the accuracy of the measurement data.

[0070] The specific steps for the denoising module 20 to correct the measurement data of the data acquisition module 1 based on the improved Extended Kalman Filter algorithm include:

[0071] S1. Obtain the signal data measured by sensors of the same type;

[0072] S2. Establish the state equation and the measurement equation, which are respectively:

[0073]

[0074] where, X k+1 is the system state matrix at time k + 1, f(V k ) is the system state transition equation at time k, W k is the system noise at time k, Z k+1 is the system measurement matrix at time k + 1, h(X k+1 ) is the system measurement equation at time k + 1, V k is the measurement noise at time k;

[0075] S3. Optimize the variables in the system noise and measurement noise at the initial moment based on the improved whale algorithm;

[0076] S4. Calculate the prior estimate value of X k ;

[0077] S5. Calculate the prior error covariance matrix;

[0078] S6. Calculate the Kalman gain;

[0079] S7. Calculate the posterior estimate value of X k ;

[0080] S8. Update the error covariance matrix.

[0081] Furthermore, the specific steps for optimizing the variables in the system noise and measurement noise at the initial moment based on the improved whale algorithm in S3 are as follows:

[0082] S31. Initialize parameters: Set the number of whale populations as N, the maximum number of iterations as T, the spatial dimension as D, and the position vector of the whale represents the variables in the system noise and measurement noise;

[0083] S32. Process boundary conditions, process each individual in the population respectively, and calculate the coefficient vectors and Generate a uniformly distributed decision random number ρ,

[0084]

[0085] where, is the control vector, and it decreases from 2 to 0 with the increase of the number of iterations throughout the iterative cycle, is a random vector on [0, 1];

[0086] S33. Calculate the fitness value: Use the mean square error value between the actual measurement value and the filtered output value as the fitness function;

[0087] S34. Predation to find the optimal solution:

[0088] Randomly search for prey: When ρ < 0.5 and , introduce a feedback mechanism,

[0089]

[0090]

[0091] where, t is the current iteration number, is the position vector of the whale individual randomly selected from the population at the t-th iteration, f rand is the corresponding whale fitness value, is the position vector of the whale individual selected by combining the optimal position in the population at the t-th iteration, α is the adjustment coefficient, β is the diversity coefficient, is the optimal position vector of the objective function in the current group, f best is the corresponding whale fitness value, f n is the fitness value corresponding to the n-th whale in the current population, is the average fitness value of the current population, represents the position of the current whale individual at the (t + 1)-th iteration, is the position vector of the current whale individual in the t-th generation. The diversity coefficient β is reflected by the fitness variance. When the diversity coefficient β remains unchanged, f rand is farther away from f best , the adjustment coefficient α is larger, making the movement amplitude of the individuals far from the optimal individual towards the optimal individual direction larger, facilitating the convergence of individuals to the optimal individual. When f rand remains unchanged, the larger the diversity coefficient β, the greater the fitness difference among individuals in the population, the higher the diversity, and the larger the adjustment coefficient α, making the individuals in the population move towards the optimal individual, using the existing information to accelerate the convergence speed;

[0092] Shrink and surround the prey: When ρ < 0.5 and , introduce an adaptive weight,

[0093]

[0094] where γ is the adaptive weight, γ0 is the set initial weight, and τ is the adjustment parameter;

[0095] Spiral update the position: When ρ ≥ 0.5,

[0096]

[0097] where s is a constant used to define the logarithmic spiral shape, and l is a uniformly distributed random number between [-1, 1];

[0098] S35. Judge the algorithm end condition, and end when the maximum number of iterations is reached or the results are the same after a certain number of iterations.

[0099] In this embodiment, the information entropy is an index used to measure the uncertainty of a random variable. In a multi-sensor system, the measurement value of each sensor can be regarded as a random variable. If the measurement value distribution of a sensor is relatively concentrated, that is, the uncertainty is small, then its information entropy is relatively low; on the contrary, if the measurement value distribution is relatively dispersed and the uncertainty is large, the information entropy is relatively high. When the information entropy of a sensor is low, it indicates that the uncertainty of its measurement value is small, meaning that the measurement result of this sensor is relatively stable and reliable. The weighted average module 21 weights the data transmitted by the sensors of the same type based on the information entropy. The specific steps include:

[0100] A1. Let n r pieces of data measured by A r sensors be transmitted to the weighted average module 21 after being corrected by the denoising module 20;

[0101] A2. For each A r sensor, count the probability that its corrected value appears within h time. Let the corrected value of the i-th A rThe correction values of the sensors have q different values, denoted as v i1 , v i2 , …, v ij , … v iq , and the corresponding frequencies are f i1 , f i2 , …, f ij , …, f iq , and the corresponding probability distribution is p i1 , p i2 , …, p ij , … p iq , then

[0102]

[0103] A3. Calculate the information entropy of each A r sensor according to the probability distribution, then

[0104]

[0105] where H(i) is the information entropy of the i-th A r sensor;

[0106] A4. Normalize the information entropy to obtain the credibility index, then

[0107]

[0108] where C(i) is the credibility index of the i-th A r sensor;

[0109] A5. Perform weighted average to determine the weight of each A r sensor according to the credibility index, then

[0110]

[0111] where w i is the weight of the i-th A r sensor;

[0112] A6. Weightedly calculate the corrected data of n r A r sensors. The weights determined by the information entropy can reflect the credibility of each sensor. Sensors with high credibility are given larger weights and contribute more to the final result in the weighted average; sensors with low credibility are given smaller weights and have less impact on the final result.

[0113] In this way, the importance of sensors in the fusion result can be adjusted according to their actual performance, making the final result more accurate and reliable.

[0114] In this embodiment, after the weighted average module 21 calculates the measurement data of sensors of the same type through weighted calculation, the data fusion module 3 fuses the data of different types of sensors based on a random forest and outputs an alarm level. The alarm level can generally be divided into three levels: safe, warning, and dangerous, and can also be specifically set to meet different environments and requirements.

[0115] Specifically, for the data of different types of sensors, each sensor may be sensitive to different features or patterns. By integrating multiple decision trees, various complex relationships and patterns in the data of different sensors can be captured, improving the prediction ability for the target variable. The random forest can fuse the data of different types of sensors to provide more comprehensive information, thereby improving the prediction accuracy. Different types of sensors may measure different physical quantities or features, and there may be complementary relationships between them. By fusing the data of these sensors through a random forest, the advantages of different sensors can be fully utilized to capture more information and improve the prediction performance of the model. The specific steps for the data fusion module 3 to fuse the data of different types of sensors based on a random forest are as follows:

[0116] B1. Use historical data as the data set E, divide the data set into a training set E1 and a test set E2. The measurement data of each type of sensor, after being corrected by the denoising module 20 and calculated by the weighted average module 21, are used as the attributes of the input data, and the alarm level is used as the output data.

[0117] B2. Construct a random forest. Use the Bootstrap algorithm to draw e samples from the training set with replacement to generate K decision trees. When splitting each node of the decision tree, randomly select m attributes from all M attributes, and select the optimal attribute from the m attributes as the splitting object according to the Gini index, and use the classification and regression tree algorithm to construct the decision tree.

[0118] B3. Use the training set data to train the random forest model.

[0119] B4. Use the test set data to evaluate the trained random forest model, and optimize the random forest model according to the evaluation results.

[0120] In this embodiment, the data transmission module 4 is wirelessly connected to an external server and can use wireless communication modules such as 4G modules, 5G modules, or WIFI modules. The data transmitted by the data transmission module 4 includes various gas concentration thresholds, control instructions, environmental parameters, alarm levels, and various gas concentrations. The control instructions can be instructions such as start detection instructions, stop detection instructions, calibration instructions, parameter setting instructions, and self-check instructions. The environmental parameters include information such as temperature, humidity, and location.

[0121] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. The IoT gas detector component valve based on integrated multi-sensor technology is characterized by: include: Data acquisition module (1): the data acquisition module (1) comprises a plurality of sensors of different types, wherein the same type of sensor comprises a plurality of sensors installed at different positions; Data processing module (2): the data processing module (2) comprises a plurality of denoising modules (20) and a plurality of weighted averaging modules (21), sensors of the same type are simultaneously connected to the same denoising module (20) by signal, and the denoising module (20) is connected to the weighted averaging module (21) by signal in a one-to-one manner; Data fusion module (3): the data fusion module (3) is simultaneously connected to the signals of a plurality of weighted average modules (21); Data transmission module (4): the data transmission module (4) is signal-connected to the data fusion module (3).

2. The IoT gas detector assembly valve based on integrated multi-sensor technology according to claim 1 is characterized in that: The specific steps of the denoising module (20) correcting the data measured by the data acquisition module (1) based on the improved extended Kalman filter algorithm include: S1. Obtain signal data measured by sensors of the same type; S2. Establish the state equation and measurement equation, which are: Among them, X k+1 is the system state matrix at time k+1, f(X k ) is the system state transfer equation at time k, W k is the system noise at time k, Z k+1 is the system measurement matrix at time k+1, h(X k+1 ) is the system measurement equation at time k+1, V k is the measurement noise at time k; S3, optimizing the variables in the system noise and measurement noise at the initial moment based on the improved whale algorithm; S4. Calculate X k A priori estimate of ; S5, calculating the prior error covariance matrix; S6, calculating Kalman gain; S7. Calculate X k The posterior estimate of ; S8. Update the error covariance matrix.

3. The IoT gas detector assembly valve based on integrated multi-sensor technology according to claim 2 is characterized in that: The specific steps of optimizing the variables in the system noise and measurement noise at the initial moment based on the improved whale algorithm in S3 are: S31, initialization parameters: set the number of whale populations to N, the maximum number of iterations to T, the spatial dimension to D, and the position vector of the whale to represent the variables in the system noise and measurement noise; S32, boundary condition processing, process the individuals in the population separately, and calculate the coefficient vector and Generate uniformly distributed decision random number ρ, in, is the control vector, and decreases from 2 to 0 as the number of iterations increases during the entire iteration cycle. is a random vector on [0,1]; S33, calculating the fitness value: taking the mean square error between the actual measured value and the filtered output value as the fitness function; S34, predation to find the optimal solution: Random search for prey: When ρ < 0.5 and When the feedback mechanism is introduced, Where t is the current iteration number, is the position vector of a randomly selected whale individual in the population at the tth iteration, f rand for The corresponding whale fitness value, is the position vector of the individual whale selected after combining the optimal position in the population at the tth iteration, α is the adjustment coefficient, β is the diversity coefficient, is the optimal position vector of the objective function in the current group, f best for The corresponding whale fitness value, f n is the fitness value corresponding to the nth whale in the current population, is the average fitness value of the current population, represents the position of the current whale individual at the t+1th iteration, is the position vector of the current whale individual in the tth generation; Contraction to surround the prey: When ρ < 0.5 and When adaptive weights are introduced, Among them, γ is the adaptive weight, γ0 is the set initial weight, and τ is the adjustment parameter; Spiral update position: When ρ ≥ 0.5, Where s is a constant used to define the shape of the logarithmic spiral, and l is a uniformly distributed random number between [-1, 1]; S35, judging the algorithm termination condition, and terminating when the maximum number of iterations is reached or the results of a certain number of iterations are the same.

4. The IoT gas detector assembly valve based on integrated multi-sensor technology according to claim 3 is characterized in that: The weighted average module (21) weights the data transmitted by sensors of the same type based on information entropy, and the specific steps include: A1. Let n r A r The data measured by the sensor are respectively corrected by the denoising module (20) and then transmitted to the weighted averaging module (21); A2. For each A r Sensor, count the probability of its correction value appearing within h time, let the i-th A r The correction value of the sensor has q different values, denoted as v i1 , v i2 , …, v ij ,…v iq , and the corresponding frequencies are f i1 , f i2 , …, f ij , …, f iq , the corresponding probability distribution is p i1 , p i2 ,…,p ij ,…p iq ,but A3. Calculate each A according to the probability distribution r The information entropy of the sensor is Among them, H(i) is the i-th A r Information entropy of the sensor; A4. Normalize the information entropy to obtain the credibility index, then Among them, C(i) is the i-th A r Sensor credibility indicators; A5, weighted average, determine each A based on the credibility index r The weight of the sensor is Among them, w i is the i-th A r The weight of the sensor; A6. Weighted calculation n r A r Sensor corrected data.

5. The IoT gas detector assembly valve based on integrated multi-sensor technology according to claim 1 is characterized in that: After the weighted average module (21) weightedly calculates the measurement data of sensors of the same type, the data fusion module (3) fuses the sensor data of different types based on random forest and outputs an alarm level.

6. The IoT gas detector assembly valve based on integrated multi-sensor technology according to claim 5 is characterized in that: The specific steps of the data fusion module (3) for fusing different types of sensor data based on random forest are as follows: B1, using historical data as a data set E, dividing the data set into a training set E1 and a test set E2, using the measurement data of each type of sensor as the attribute of the input data after being corrected by the denoising module (20) and calculated by the weighted average module (21), and using the alarm level as the output data; B2. Construct a random forest. Use the Bootstrap algorithm to extract e samples with replacement from the training set to generate K decision trees. When splitting each node of the decision tree, randomly select m attributes from all M attributes. Select the best attribute from the m attributes as the splitting object according to the Gini index. Use the classification and regression tree algorithm to construct the decision tree. B3. Use the training set data to train the random forest model; B4. Use the test set data to evaluate the trained random forest model and optimize the random forest model based on the evaluation results.

7. The IoT gas detector assembly valve based on integrated multi-sensor technology according to claim 1 is characterized in that: The data transmission module (4) is connected to an external server via wireless signals, and the data transmission module (4) transmits data including various gas concentration thresholds, control instructions, environmental parameters, alarm levels and various gas concentrations.

Images