Fumigation treatment multi-parameter comprehensive monitoring system

Abstract

The invention discloses a fumigation treatment multi-parameter comprehensive monitoring system which comprises a multi-parameter sensor monitoring module, a multi-modal data processing and analyzing module, a system intelligent monitoring and adjusting module, a fumigation environment control module, a remote maintenance and management module and a system interaction and feedback module. The multi-parameter sensor monitoring module is used for collecting parameter data. The multi-modal data processing and analyzing module is used for fusing and analyzing data. The system intelligent monitoring and adjusting module is used for intelligently monitoring and adjusting fumigation strategies. The fumigation environment control module is used for adjusting environment parameters. The remote maintenance and management module is used for system maintenance and management. And the system interaction and feedback module is used for display interaction of data and a control state. According to the fumigation treatment multi-parameter comprehensive monitoring system, a multi-modal data fusion algorithm based on information channel decomposition is provided for multi-parameter data fusion and analysis; and a multi-parameter intelligent monitoring algorithm based on a multi-layer feedforward neural network is provided to intelligently monitor the fumigation environment.

Description

Technical Field

[0001] The invention relates to the fields of multimodal data fusion, multi-layer feedforward neural network and multi-parameter monitoring, and in particular to a fumigation treatment multi-parameter comprehensive monitoring system. Background Art

[0002] Multimodal data fusion technology is a technology that integrates multimodal data from different sensors. It aims to solve the limitations of data collected by a single sensor. Traditional fumigation treatment systems usually rely on a single type of sensor for data collection. However, due to the complexity and diversity of environmental factors, the data from a single sensor often cannot fully and accurately reflect the true state of the fumigation environment. Multimodal data fusion technology effectively integrates multiple data sources to ensure that various environmental parameters in the fumigation process can be comprehensively evaluated, thereby improving the accuracy and reliability of the data and ensuring that various key parameters in the fumigation process are monitored comprehensively and in real time.

[0003] Multilayer feedforward neural network technology is a deep learning method that processes input data through a multilayer neural network structure to achieve automatic learning and recognition of complex patterns. Multilayer feedforward neural network technology can be used to intelligently analyze multi-parameter data in the fumigation environment, and then predict and optimize the environmental state. By using a multilayer feedforward neural network to conduct in-depth learning of large amounts of data, this technology can extract valuable information from complex multidimensional data, automatically identify potential abnormal situations, and provide real-time feedback for system regulation. This technology can improve the intelligence and response speed of fumigation environment monitoring and ensure precise control of the fumigation process.

[0004] However, the existing multi-parameter comprehensive monitoring system for fumigation treatment has the problem of inaccurate data and missing key parameters due to the data collection method using a single sensor that cannot fully cover the diversity of the environment. Secondly, traditional monitoring methods are difficult to achieve real-time intelligent analysis of complex environments, resulting in low accuracy and efficiency of fumigation treatment, and difficulty in timely adjusting control strategies. Summary of the invention

[0005] The purpose of the present invention is to provide a multi-parameter comprehensive monitoring system for fumigation treatment to solve the problems of inaccurate data and missing key parameters in the existing multi-parameter comprehensive monitoring system for fumigation treatment proposed in the above background technology, because the data collection method using a single sensor cannot fully cover the diversity of the environment, secondly, the traditional monitoring method is difficult to achieve real-time intelligent analysis of complex environments, resulting in low accuracy and efficiency of fumigation treatment, and it is difficult to adjust the control strategy in time.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a fumigation treatment multi-parameter comprehensive monitoring system, including a multi-parameter sensor monitoring module, a multi-modal data processing and analysis module, a system intelligent monitoring and adjustment module, a fumigation environment control module, a remote maintenance and management module, and a system interaction and feedback module, characterized in that: the multi-parameter sensor monitoring module is used to collect multi-dimensional parameter data in the fumigation environment in real time, including fumigation gas concentration, temperature and humidity, and pressure, to ensure comprehensive monitoring of the environmental state and real-time data updating; the multi-modal data processing and analysis module includes a data anomaly detection unit and a multi-modal data fusion unit, the data anomaly detection unit is used to identify and mark abnormal values ​​and noise in the monitoring data to ensure the accuracy and stability of data input, and the multi-modal data fusion unit proposes a multi-modal data fusion algorithm based on information channel decomposition, which is used to perform feature extraction, comprehensive analysis and information integration on multi-source monitoring data, to ensure the effective coordination and accurate interpretation of multi-parameter data, and to provide reliable intelligent monitoring and regulation of the fumigation environment. Data support; the system intelligent monitoring and regulation module includes an intelligent comprehensive monitoring unit and a dynamic feedback control unit. The intelligent comprehensive monitoring unit proposes a multi-parameter intelligent monitoring algorithm based on a multi-layer feedforward neural network to perform real-time analysis and status evaluation on the key parameters of temperature, humidity, gas concentration, and pressure in the fumigation environment after being processed by the multimodal data processing and analysis module, ensuring a comprehensive perception of the fumigation process and an accurate judgment of the environmental status. The dynamic feedback control unit is used to regulate and control the delivery and concentration adjustment of the fumigation agent to ensure the effect and safety of the fumigation treatment; the fumigation environment control module is used to adjust and control the key parameters of the environment during the fumigation process, including ambient temperature, humidity, and pressure parameters, to ensure the uniformity and efficiency of the fumigation process; the remote maintenance and management module is used to realize remote monitoring, maintenance, and parameter management of the system to ensure the efficiency of equipment operation; the system interaction and feedback module is used to provide display of real-time monitoring data, regulation status, and abnormal alarms to ensure the operability of the system and the convenience of user experience.

[0007] Preferably, the multi-parameter sensor monitoring module monitors key parameters in the fumigation environment in real time through multiple sensors, including gas concentration, temperature and humidity, and pressure parameters, to ensure that the environmental status can be fully tracked and real-time data can be provided, providing accurate data support for subsequent processing and intelligent regulation.

[0008] Preferably, the multimodal data processing and analysis module includes a data anomaly detection unit, which detects abnormal changes, including drastic fluctuations in gas concentration and abnormal changes in temperature and humidity, by performing real-time analysis on the monitoring data, to ensure that problems can be discovered as early as possible and the alarm system can be triggered.

[0009] Preferably, the multimodal data processing and analysis module includes a multimodal data fusion unit, which proposes a multimodal data fusion algorithm based on information channel decomposition to fuse and analyze data from different sensors, including gas concentration, temperature and humidity, and pressure, by decomposing the signal characteristics of each information channel, thereby ensuring that the system can comprehensively and accurately obtain the status information of the fumigation environment and provide effective support for subsequent intelligent control.

[0010] Specifically, the multimodal data fusion algorithm based on information channel decomposition is as follows: First, the multimodal data obtained from the multi-parameter sensor monitoring module is divided into two types of modes, including the gas concentration of fumigation treatment and the temperature, humidity and pressure data of the fumigation environment, respectively, to form two modal data sets, namely the gas concentration set X and the temperature and humidity set Y. The multimodal data is fused based on the correlation between the two types of data, and an information channel is constructed to quantify the degree of information transmission between the two types of data. The specific formula is expressed as follows:

[0011]

[0012] Among them, P(Y|X) represents the conditional probability matrix of the first information channel, p(y i |x j ) represents the gas concentration data set X for fumigation treatment, with the value x j The temperature, humidity and pressure data set Y is y i The conditional probability of, i represents the number index of temperature, humidity and pressure data sets, j represents the number index of fumigation gas concentration data sets, n(x j ,y i ) is expressed as the gas concentration that satisfies the fumigation treatment at the same time is x j And the temperature, humidity and pressure data are y i The number of samples, n(x) represents the total number of samples with a gas concentration of x in all fumigation treatments, P(X|Y) represents the conditional probability matrix of the second information channel, p(x j |y i ) represents the temperature, humidity and pressure data set Y with the value of y i The gas concentration data set Y of the fumigation treatment is taken as y i The conditional probability of n(y) is expressed as the total number of samples with temperature, humidity and pressure data taking the value of y. Then, the marginal probability distribution P(X) and P(Y) of each sensor data are calculated, and the mutual information I(X; Y) of X and Y is quantified by the distribution. The mutual information is used to measure the amount of information sharing between sensor data. The specific formula is expressed as follows:

[0013]

[0014] Among them, p(x j ) and p(y i ) represents the marginal probability of X and Y, N represents the total number of multimodal data samples, n(x j ) represents the gas concentration of fumigation treatment, which is x j The total number of samples, n(y i ) represents the temperature, humidity and pressure data value is y i The total number of samples, I(X;Y) is used to quantify the amount of information shared between two data sets X and Y, p(x j ,y i ) represents the value x in the set X j And the set Y takes the value y i The joint probability of simultaneous occurrence, log, is expressed as a logarithmic function. If p(x j ,y i )>p(x j )p(y i ), it means that there is information association between the multimodal data of set X and set Y. If p(x j ,y i )≤p(x j )p(y i ), it means that the multimodal data between set X and set Y are independent of each other.

[0015]

[0016]

[0017] Among them, I spec (x j ; Y) represents the data point x j The specific information contribution to the set Y, I spec (y i ; X) represents the data point y i For the specific information contribution of set X, by calculating the specific information contribution of two modal data information to the other modal data information respectively, the redundant information influence of the data point is eliminated, so as to ensure that each sensor data point can accurately reflect its environmental uniqueness in the fusion. Based on the calculated specific information amount, the value with the largest amount of information is selected in each pair of corresponding sensor data points as part of the fusion result to achieve point-by-point optimization fusion of multimodal data. The specific formula is expressed as:

[0018]

[0019] Among them, z kIt is represented as the kth data information of the fused data set, and k is represented as the specific data point position index currently being compared and fused. By comparing specific information, it is ensured that the fused data points in the fumigation process reflect the key information characteristics of the environment to the greatest extent. Finally, all fused data points are combined into a complete data set Z, and the entropy of the fused data set is calculated to quantify the multimodal information gain and ensure the effectiveness of the fusion result. The specific formula is expressed as:

[0020]

[0021] Among them, H(Z) represents the entropy of the fused data set Z, z k It is represented as the kth data value after fusion, p(z k ) is represented as data point z k The probability distribution of multimodal data is quantified by the overall information gain after multimodal data fusion, which can ensure the effect and accuracy of the fusion algorithm. By calculating the entropy of the fused data set, the multimodal data fusion algorithm based on information channel decomposition quantifies the overall information gain after multimodal data fusion in the fumigation environment, further ensuring the effectiveness of the algorithm.

[0022] Preferably, the system intelligent monitoring and regulation module includes an intelligent comprehensive monitoring unit, which proposes a multi-parameter intelligent monitoring algorithm based on a multi-layer feedforward neural network to perform deep learning and processing on data from sensors, ensuring that each parameter of the fumigation environment can be monitored in real time and accurately.

[0023] Specifically, the multi-parameter intelligent monitoring algorithm based on the multi-layer feedforward neural network is as follows: First, according to the data information obtained from the multi-modal data processing and analysis module after processing, a multi-layer feedforward neural network model is constructed, and the input layer, hidden layer and output layer of the neural network are defined. The input layer nodes represent the fumigation environment parameters processed by the multi-modal data processing and analysis module, including gas concentration, temperature, humidity and pressure. The hidden layer nodes are used to learn the complex nonlinear relationship of the input data, and the output layer nodes predict the trend of key parameters of the fumigation environment. The connection weights and bias thresholds of each layer of nodes are assigned initial values ​​to ensure that the neural network has a good initial state during training. The specific formula is expressed as follows:

[0024]

[0025] b j =0

[0026] Among them, w ij is the weight connecting the input node i and the hidden node j, n is the number of input layer nodes, b jIt is represented as the bias of hidden node j, i is represented as the input node number index, j is represented as the hidden node number index, and U is represented as the uniform distribution function. By initializing the weights and thresholds, it provides a starting point for training for the multi-layer feedforward neural network, ensuring the randomness of the weights to avoid falling into the local optimum. Then, the input parameter data is passed to the hidden layer and the output layer in turn, and the activation value and output value of each node are calculated. The output function of the hidden layer is specifically expressed as:

[0027]

[0028] Among them, h j is the output value of the jth node in the hidden layer, σ is the activation function of the hidden layer, which uses a nonlinear function to capture the complex features of the input data, I is the total number of hidden nodes, and x is i The input value of the i-th node in the input layer, e, is expressed as an exponential function, and the output function of the output layer is specifically expressed as:

[0029]

[0030] Among them, k It represents the output value of the kth node in the output layer, reflecting the prediction result of the multi-layer feedforward neural network on the input data. k represents the index of the number of output nodes, H represents the index of the number of hidden layer nodes, and v jk It is represented as the connection weight between the jth node in the hidden layer and the kth node in the output layer, h j It is expressed as the output value of the jth node in the hidden layer, which represents the value obtained after the hidden layer processes the input through the activation function, reflecting the nonlinear transformation and feature extraction capabilities of the hidden layer nodes for the input data. k It is expressed as the bias of the kth node in the output layer, which is used to adjust the activation level of the node and improve the fitting ability of the neural network. It generates the predicted value o through forward propagation calculation. k , used to evaluate the parameter change trend of the fumigation environment and provide a basis for real-time intelligent monitoring. Secondly, the error between the output value and the target value is calculated, and the weight and threshold are adjusted through the back propagation algorithm. The error propagation is divided into the gradient update process of the output layer and the hidden layer. The purpose is to minimize the prediction error. The specific formula for calculating the output layer error is expressed as:

[0031] δ k =(t k -o k ) k (1-o k )

[0032] Among them, δ k Represented as the error signal of the kth neuron in the output layer, used to update the adjustment amount of the neuron and its associated weights, t kIt is expressed as the target output value of the kth neuron in the output layer. The specific formula for calculating the hidden layer error is expressed as:

[0033]

[0034] Among them, Δ j It is represented as the error gradient of the jth neuron in the hidden layer, reflecting the contribution of the current hidden neuron to the overall error. O is represented as the total number of output nodes. The specific formula for weight update calculation is expressed as:

[0035] w ij (n) = w ij (n-1)+ηδ j x i ,v jk (n) = v jk (n)+ηδ k h j

[0036] Among them, η represents the learning step size, n represents the iteration number index, and finally, the forward propagation and back propagation steps are repeated until the network output error meets the set accuracy requirement, that is, the maximum number of training times is reached. The specific formula is expressed as:

[0037]

[0038] Where E represents the objective function for calculating the overall error in the multi-layer feedforward neural network, t k It is represented as the target value of the output node k, and ε is represented as the set error accuracy. Through multiple iterative training, the neural network gradually learns the complex relationship of multiple parameters in the fumigation environment, realizes accurate monitoring and trend prediction, and provides strong support for the intelligent monitoring of the system.

[0039] Preferably, the system intelligent monitoring and regulation module includes a dynamic feedback regulation unit, which regulates and controls the delivery and concentration of the fumigation agent according to the analysis results of the intelligent comprehensive monitoring unit to ensure the optimized stability of the fumigation effect. Through the feedback regulation strategy, it is ensured that the system responds to the monitored changes in real time and optimizes the fumigation process.

[0040] Preferably, the fumigation environment control module ensures that the fumigation environment can be maintained in an ideal state by adjusting and controlling key parameters of the environment during the fumigation process, including ambient temperature, humidity and pressure parameters.

[0041] Preferably, the remote maintenance and management module realizes remote access, monitoring and management of the entire system through remote communication technology, supports remote maintenance functions such as equipment status monitoring, troubleshooting, and updates and upgrades, and ensures the stability and long-term operation of the system.

[0042] Preferably, the system interaction and feedback module provides real-time data display, parameter adjustment function and fault warning information of the fumigation process by constructing an interactive interface, ensuring that the user can intuitively understand the fumigation process and make real-time adjustments.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. Multimodal data fusion unit A multimodal data fusion algorithm based on information channel decomposition is proposed. The data obtained from multiple sensors, including gas concentration, temperature, humidity and pressure data of fumigation treatment, are effectively combined through a multimodal data fusion algorithm based on information channel decomposition, so as to optimize the control and monitoring of the fumigation process. First, the multimodal data fusion algorithm based on information channel decomposition quantifies the information transmission relationship between the fumigation gas concentration data set and the environmental temperature, humidity and pressure data set by constructing information channels, and uses conditional probability matrices to describe the dependencies between the data. Specifically, the multimodal data fusion algorithm based on information channel decomposition measures the changes of one modal data when another modal data is known by calculation, providing a clear probabilistic framework for data fusion. Through the construction of this information channel, the system can effectively capture and quantify the mutual relationship and information sharing between the two modal data, ensuring that the various monitoring data of the fumigation treatment environment can accurately reflect the actual environmental status. Then, the algorithm calculates the marginal probability distribution of each sensor data, and uses mutual information to further quantify the shared information between the fumigation gas concentration and the temperature, humidity and pressure data. Information, through the calculation of mutual information, the algorithm can measure the correlation between different sensor data, help the system identify sensor data with more information value for optimizing the fumigation environment, enhance the understanding of the correlation between different sensor data, enable the system to better fuse various monitoring data, and avoid the interference of information redundancy. More importantly, based on the decomposition of mutual information, the algorithm further refines the contribution of each data point in the fusion by calculating the specific information of multimodal data. This step ensures that each sensor data point accurately reflects its uniqueness in the fusion by removing redundant information. The algorithm can realize point-by-point optimization fusion of multimodal data, maximize the retention of the key features of each sensor data, and enhance the accuracy and responsiveness of the system. Finally, by calculating the entropy of the fused data set, the multimodal data fusion algorithm based on information channel decomposition quantifies the overall information gain after the fusion of multimodal data in the fumigation environment, further ensuring the effectiveness of the algorithm. The calculation of entropy can ensure the rationality and consistency of the fusion results, avoid information loss and error accumulation, and ensure that the system can achieve efficient and accurate monitoring and control of the fumigation process, optimize the fumigation treatment effect and improve operational safety.

[0045] 2. The intelligent comprehensive monitoring unit proposes a multi-parameter intelligent monitoring algorithm based on a multi-layer feedforward neural network. First, a multi-layer feedforward neural network model is constructed based on the results of multimodal data processing, and the structure of the input layer, hidden layer and output layer are defined. In the input layer, multiple monitoring parameters of the fumigation environment, including key parameters of gas concentration, temperature, humidity and pressure, are used as input features. These complex nonlinear relationships are learned through the hidden layer nodes, and finally the changing trends of key parameters in the fumigation environment are predicted through the output layer. By initializing the connection weights and biases of the neural network, it is ensured that the network can effectively avoid local optimal solutions during the training process, thereby improving the stability and convergence speed of the training. During the training process, by calculating the activation value and output value of each node, the activation function of the hidden layer uses a nonlinear function to capture the complex features in the input data. In the output layer, the network generates prediction results by weighted summation, which reflects the comprehensive understanding and prediction ability of the neural network for the input data. To ensure the accuracy of the network, the back propagation algorithm is used to calculate the error between the output value and the target value, and the weights and thresholds in the neural network are adjusted by the gradient descent method. The error of the output layer reflects the difference between the target value and the predicted value, while the error of the hidden layer helps to update the weight of each hidden node, thereby further improving the learning ability of the network. Through multiple iterations of forward propagation and back propagation, the neural network can gradually learn and optimize the prediction ability of fumigation environmental parameters. Through the training and optimization of this multi-layer feedforward neural network, the system can accurately capture the complex relationship between multiple parameters in the fumigation environment and realize effective prediction of future environmental changes. In general, the neural network model can provide real-time environmental monitoring and early warning, provide strong data support for the optimization and regulation of the fumigation process, and ensure the precise control and safety of the fumigation process. The multi-parameter intelligent monitoring algorithm based on the multi-layer feedforward neural network not only improves the efficiency of fumigation treatment, but also enhances the sensitivity to changes in environmental parameters, enabling the system to respond quickly and accurately when facing complex and dynamic fumigation environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The invention is further illustrated by the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0047] Figure 1 It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] See also Figure 1 The present invention provides a multi-parameter comprehensive monitoring system for fumigation treatment, including a multi-parameter sensor monitoring module, a multi-modal data processing and analysis module, a system intelligent monitoring and adjustment module, a fumigation environment control module, a remote maintenance and management module, and a system interaction and feedback module, characterized in that: the multi-parameter sensor monitoring module is used to collect multi-dimensional parameter data in the fumigation environment in real time, including fumigation gas concentration, temperature and humidity, and pressure, to ensure comprehensive monitoring of the environmental state and real-time data updating; the multi-modal data processing and analysis module includes a data anomaly detection unit and a multi-modal data fusion unit, the data anomaly detection unit is used to identify and mark abnormal values ​​and noise in the monitoring data to ensure the accuracy and stability of data input, and the multi-modal data fusion unit proposes a multi-modal data fusion algorithm based on information channel decomposition, which is used to extract features, comprehensively analyze and integrate information from multi-source monitoring data, ensure the effective coordination and accurate interpretation of multi-parameter data, and provide reliable data support for intelligent monitoring and regulation of the fumigation environment; The intelligent monitoring and regulation module of the system includes an intelligent comprehensive monitoring unit and a dynamic feedback control unit. The intelligent comprehensive monitoring unit proposes a multi-parameter intelligent monitoring algorithm based on a multi-layer feedforward neural network to perform real-time analysis and status evaluation on the key parameters of temperature, humidity, gas concentration, and pressure in the fumigation environment after being processed by the multimodal data processing and analysis module, ensuring a comprehensive perception of the fumigation process and an accurate judgment of the environmental status. The dynamic feedback control unit is used to regulate and control the delivery and concentration of fumigation agents to ensure the effectiveness and safety of fumigation treatment; the fumigation environment control module is used to adjust and control the key parameters of the environment during the fumigation process, including ambient temperature, humidity, and pressure parameters, to ensure the uniformity and efficiency of the fumigation process; the remote maintenance and management module is used to realize remote monitoring, maintenance, and parameter management of the system to ensure the efficiency of equipment operation; the system interaction and feedback module is used to provide real-time monitoring data, regulation status, and abnormal alarms to ensure the operability of the system and the convenience of user experience.

[0050] See also Figure 1 ,Furthermore, the multi-parameter sensor monitoring module monitors the key parameters in the fumigation environment, including gas concentration, temperature and humidity, and pressure parameters, in real time through multiple sensors, ensuring that the environmental status can be fully tracked and real-time data can be provided, providing accurate data support for subsequent processing and intelligent regulation.

[0051] See also Figure 1 Furthermore, the multimodal data processing and analysis module includes a data anomaly detection unit, which detects abnormal changes, including drastic fluctuations in gas concentration and abnormal changes in temperature and humidity, by performing real-time analysis on the monitoring data, to ensure that problems can be discovered as early as possible and the alarm system can be triggered.

[0052] See also Figure 1 Furthermore, the multimodal data processing and analysis module includes a multimodal data fusion unit. The multimodal data fusion unit proposes a multimodal data fusion algorithm based on information channel decomposition to fuse and analyze data from different sensors, including gas concentration, temperature and humidity, and pressure, by decomposing the signal characteristics of each information channel, thereby ensuring that the system can comprehensively and accurately obtain the status information of the fumigation environment and provide effective support for subsequent intelligent control.

[0053] See also Figure 1 ,Furthermore, the multimodal data fusion algorithm based on information channel decomposition is specifically as follows: First, the multimodal data obtained from the multi-parameter sensor monitoring module is divided into two types of modes, including the gas concentration of fumigation treatment and the temperature, humidity and pressure data of the fumigation environment, respectively, to form two modal data sets, namely the gas concentration set X and the temperature and humidity set Y. The multimodal data is fused based on the correlation between the two types of data, and an information channel is constructed to quantify the degree of information transmission between the two types of data. The specific formula is expressed as follows:

[0054]

[0055] Among them, P(Y|X) represents the conditional probability matrix of the first information channel, p(y i |x j ) represents the gas concentration data set X for fumigation treatment, with the value x j The temperature, humidity and pressure data set Y is y i The conditional probability of, i represents the number index of temperature, humidity and pressure data sets, j represents the number index of fumigation gas concentration data sets, n(x j ,y i ) is expressed as the gas concentration that satisfies the fumigation treatment at the same time is x j And the temperature, humidity and pressure data are y i The number of samples, n(x) represents the total number of samples with a gas concentration of x in all fumigation treatments, P(X|Y) represents the conditional probability matrix of the second information channel, p(x j |y i ) represents the temperature, humidity and pressure data set Y with the value of y i The gas concentration data set Y of the fumigation treatment is taken as yi The conditional probability of n(y) is expressed as the total number of samples with temperature, humidity and pressure data taking the value of y. Then, the marginal probability distribution P(X) and P(Y) of each sensor data are calculated, and the mutual information I(X; Y) of X and Y is quantified by the distribution. The mutual information is used to measure the amount of information sharing between sensor data. The specific formula is expressed as follows:

[0056]

[0057] Among them, p(x j ) and p(y i ) represents the marginal probability of X and Y, N represents the total number of multimodal data samples, n(x j ) represents the gas concentration of fumigation treatment, which is x j The total number of samples, n(y i ) represents the temperature, humidity and pressure data value is y i The total number of samples, I(X;Y) is used to quantify the amount of information shared between two data sets X and Y, p(x j ,y i ) represents the value x in the set X j And the set Y takes the value y i The joint probability of simultaneous occurrence, log, is expressed as a logarithmic function. If p(x j ,y i )>p(x j )p(y i ), it means that there is information association between the multimodal data of set X and set Y. If p(x j ,y i )≤p(x j )p(y i ), it means that the multimodal data between set X and set Y are independent of each other.

[0058]

[0059] Among them, I spec (x j ; Y) represents the data point x j The specific information contribution to the set Y, I spec (y i ; X) represents the data point y iFor the specific information contribution of set X, by calculating the specific information contribution of two modal data information to the other modal data information respectively, the redundant information influence of the data point is eliminated, so as to ensure that each sensor data point can accurately reflect its environmental uniqueness in the fusion. Based on the calculated specific information amount, the value with the largest amount of information is selected in each pair of corresponding sensor data points as part of the fusion result to achieve point-by-point optimization fusion of multimodal data. The specific formula is expressed as:

[0060]

[0061] Among them, z k It is represented as the kth data information of the fused data set, and k is represented as the specific data point position index currently being compared and fused. By comparing specific information, it is ensured that the fused data points in the fumigation process reflect the key information characteristics of the environment to the greatest extent. Finally, all fused data points are combined into a complete data set Z, and the entropy of the fused data set is calculated to quantify the multimodal information gain and ensure the effectiveness of the fusion result. The specific formula is expressed as:

[0062]

[0063] Among them, H(Z) represents the entropy of the fused data set Z, z k It is represented as the kth data value after fusion, p(z k ) is represented as data point z k The probability distribution of multimodal data is quantified by the overall information gain after multimodal data fusion, which can ensure the effect and accuracy of the fusion algorithm. By calculating the entropy of the fused data set, the multimodal data fusion algorithm based on information channel decomposition quantifies the overall information gain after multimodal data fusion in the fumigation environment, further ensuring the effectiveness of the algorithm.

[0064] See also Figure 1 Furthermore, the intelligent monitoring and adjustment module of the system includes an intelligent comprehensive monitoring unit. The intelligent comprehensive monitoring unit proposes a multi-parameter intelligent monitoring algorithm based on a multi-layer feedforward neural network to perform deep learning and processing on the data from the sensor, ensuring that each parameter of the fumigation environment can be monitored in real time and accurately.

[0065] See also Figure 1Furthermore, the multi-parameter intelligent monitoring algorithm based on the multi-layer feedforward neural network is specifically as follows: First, according to the data information obtained from the multi-modal data processing and analysis module after processing, a multi-layer feedforward neural network model is constructed, and the input layer, hidden layer and output layer of the neural network are defined. The input layer nodes represent the fumigation environment parameters processed by the multi-modal data processing and analysis module, including gas concentration, temperature, humidity and pressure. The hidden layer nodes are used to learn the complex nonlinear relationship of the input data, and the output layer nodes predict the trend of key parameters of the fumigation environment. The connection weights and bias thresholds of each layer of nodes are assigned initial values ​​to ensure that the neural network has a good initial state during training. The specific formula is expressed as follows:

[0066]

[0067] Among them, w ij is the weight connecting the input node i and the hidden node j, n is the number of input layer nodes, b j It is represented as the bias of hidden node j, i is represented as the input node number index, j is represented as the hidden node number index, and U is represented as the uniform distribution function. By initializing the weights and thresholds, it provides a starting point for training for the multi-layer feedforward neural network, ensuring the randomness of the weights to avoid falling into the local optimum. Then, the input parameter data is passed to the hidden layer and the output layer in turn, and the activation value and output value of each node are calculated. The output function of the hidden layer is specifically expressed as:

[0068]

[0069] Among them, h j is the output value of the jth node in the hidden layer, σ is the activation function of the hidden layer, which uses a nonlinear function to capture the complex features of the input data, I is the total number of hidden nodes, and x is i The input value of the i-th node in the input layer, e, is expressed as an exponential function, and the output function of the output layer is specifically expressed as:

[0070]

[0071] Among them, k It represents the output value of the kth node in the output layer, reflecting the prediction result of the multi-layer feedforward neural network on the input data. k represents the index of the number of output nodes, H represents the index of the number of hidden layer nodes, and v jk It is represented as the connection weight between the jth node in the hidden layer and the kth node in the output layer, h j It is expressed as the output value of the jth node in the hidden layer, which represents the value obtained after the hidden layer processes the input through the activation function, reflecting the nonlinear transformation and feature extraction capabilities of the hidden layer nodes for the input data. kIt is expressed as the bias of the kth node in the output layer, which is used to adjust the activation level of the node and improve the fitting ability of the neural network. It generates the predicted value o through forward propagation calculation. k , used to evaluate the parameter change trend of the fumigation environment and provide a basis for real-time intelligent monitoring. Secondly, the error between the output value and the target value is calculated, and the weight and threshold are adjusted through the back propagation algorithm. The error propagation is divided into the gradient update process of the output layer and the hidden layer. The purpose is to minimize the prediction error. The specific formula for calculating the output layer error is expressed as:

[0072] δ k =(t k -o k ) k (1-o k )

[0073] Among them, δ k Represented as the error signal of the kth neuron in the output layer, used to update the adjustment amount of the neuron and its associated weights, t k It is expressed as the target output value of the kth neuron in the output layer. The specific formula for calculating the hidden layer error is expressed as:

[0074]

[0075] Among them, Δ j It is represented as the error gradient of the jth neuron in the hidden layer, reflecting the contribution of the current hidden neuron to the overall error. O is represented as the total number of output nodes. The specific formula for weight update calculation is expressed as:

[0076] w ij (n) = w ij (n-1)+ηδ j x i ,v jk (n) = v jk (n)+ηδ k h j

[0077] Among them, η represents the learning step size, n represents the iteration number index, and finally, the forward propagation and back propagation steps are repeated until the network output error meets the set accuracy requirement, that is, the maximum number of training times is reached. The specific formula is expressed as:

[0078]

[0079] Where E represents the objective function for calculating the overall error in the multi-layer feedforward neural network, t kIt is represented as the target value of the output node k, and ε is represented as the set error accuracy. Through multiple iterative training, the neural network gradually learns the complex relationship of multiple parameters in the fumigation environment, realizes accurate monitoring and trend prediction, and provides strong support for the intelligent monitoring of the system.

[0080] Preferably, the system intelligent monitoring and regulation module includes a dynamic feedback regulation unit, which regulates and controls the delivery and concentration of the fumigation agent according to the analysis results of the intelligent comprehensive monitoring unit to ensure the optimized stability of the fumigation effect. Through the feedback regulation strategy, it is ensured that the system responds to the monitored changes in real time and optimizes the fumigation process.

[0081] See also Figure 1 Furthermore, the fumigation environment control module ensures that the fumigation environment can be maintained in an ideal state by adjusting and controlling key parameters of the environment during the fumigation process, including ambient temperature, humidity and pressure parameters.

[0082] See also Figure 1 Furthermore, the remote maintenance and management module realizes remote access, monitoring and management of the entire system through remote communication technology, supports remote maintenance functions such as equipment status monitoring, troubleshooting, and updates and upgrades, and ensures the stability and long-term operation of the system.

[0083] See also Figure 1 ,Furthermore, the system interaction and feedback module provides real-time data display, ,parameter adjustment function and fault warning information of the fumigation process by constructing ,an interactive interface, ensuring that the user can intuitively understand the fumigation process and make ,real-time adjustments.

[0084] When used specifically, first, the multi-parameter sensor monitoring module is used to collect multi-dimensional parameter data in the fumigation environment in real time, including fumigation gas concentration, temperature, humidity, and pressure, to ensure comprehensive monitoring of the environmental status and real-time data updating; secondly, the multi-modal data processing and analysis module includes a data anomaly detection unit and a multi-modal data fusion unit. The data anomaly detection unit is used to identify and mark abnormal values ​​and noise in the monitoring data to ensure the accuracy and stability of data input. The multi-modal data fusion unit proposes a multi-modal data fusion algorithm based on information channel decomposition, which is used to extract features, comprehensively analyze and integrate information from multi-source monitoring data to ensure effective coordination and accurate interpretation of multi-parameter data, and provide reliable data support for intelligent monitoring and regulation of the fumigation environment; then, the system intelligent monitoring and regulation module includes an intelligent comprehensive monitoring unit and a dynamic feedback regulation unit. The intelligent comprehensive monitoring unit proposes a multi-modal data fusion algorithm based on information channel decomposition. The multi-parameter intelligent monitoring algorithm of the multi-layer feedforward neural network performs real-time analysis and status evaluation on the key parameters of temperature, humidity, gas concentration, and pressure in the fumigation environment after being processed by the multimodal data processing and analysis module, ensuring a comprehensive perception of the fumigation process and an accurate judgment of the environmental status. The dynamic feedback control unit is used to regulate and control the delivery and concentration of fumigation agents to ensure the effectiveness and safety of the fumigation treatment. Secondly, the fumigation environment control module is used to adjust and control the key parameters of the environment during the fumigation process, including ambient temperature, humidity, and pressure parameters, to ensure the uniformity and efficiency of the fumigation process. Finally, the remote maintenance and management module is used to realize remote monitoring, maintenance, and parameter management of the system to ensure the efficiency of equipment operation. The system interaction and feedback module is used to provide real-time monitoring data, control status, and abnormal alarms to ensure the operability of the system and the convenience of user experience.

[0085] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-parameter integrated monitoring system for fumigation treatment, comprising a multi-parameter sensor monitoring module, a multi-modal data processing and analysis module, a system intelligent monitoring and adjustment module, a fumigation environment control module, a remote maintenance and management module, and a system interaction and feedback module, characterized in that: The multi-parameter sensor monitoring module is used to collect multi-dimensional parameter data in the fumigation environment in real time, including fumigation gas concentration, temperature, humidity, and pressure, to ensure comprehensive monitoring of the environmental status and real-time data updating; the multi-modal data processing and analysis module includes a data anomaly detection unit and a multi-modal data fusion unit. The data anomaly detection unit is used to identify and mark abnormal values ​​and noise in the monitoring data to ensure the accuracy and stability of data input. The multi-modal data fusion unit proposes a multi-modal data fusion algorithm based on information channel decomposition, which is used to extract features, comprehensively analyze and integrate information from multi-source monitoring data to ensure effective coordination and accurate interpretation of multi-parameter data, and provide reliable data support for intelligent monitoring and regulation of the fumigation environment; the system intelligent monitoring and regulation module includes an intelligent comprehensive monitoring unit and a dynamic feedback regulation unit. The intelligent comprehensive monitoring unit proposes a multi-modal data fusion algorithm based on multi-layer feedforward neural network. The multi-parameter intelligent monitoring algorithm through the network performs real-time analysis and status evaluation on the key parameters of temperature, humidity, gas concentration and pressure in the fumigation environment after being processed by the multimodal data processing and analysis module, ensuring the comprehensive perception of the fumigation process and the accurate judgment of the environmental status. The dynamic feedback control unit is used to adjust and control the delivery and concentration adjustment of the fumigation agent to ensure the effect and safety of the fumigation treatment; the fumigation environment control module is used to adjust and control the key parameters of the environment during the fumigation treatment, including the ambient temperature, humidity and pressure parameters, to ensure the uniformity and efficiency of the fumigation process; the remote maintenance and management module is used to realize remote monitoring, maintenance and parameter management of the system to ensure the efficiency of equipment operation; the system interaction and feedback module is used to provide display of real-time monitoring data, control status and abnormal alarms to ensure the operability of the system and the convenience of user experience.

2. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 1, characterized in that: The multi-parameter sensor monitoring module monitors key parameters in the fumigation environment in real time through multiple sensors, including gas concentration, temperature and humidity, and pressure parameters, ensuring that the environmental status can be fully tracked and real-time data can be provided, providing accurate data support for subsequent processing and intelligent regulation.

3. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 1, characterized in that: The multimodal data processing and analysis module includes a data anomaly detection unit and a multimodal data fusion unit. The data anomaly detection unit detects abnormal changes, including drastic fluctuations in gas concentration and abnormal changes in temperature and humidity, by performing real-time analysis on the monitoring data, to ensure that problems can be discovered as early as possible and the alarm system can be triggered. The multimodal data fusion unit proposes a multimodal data fusion algorithm based on information channel decomposition to fuse and analyze data from different sensors, including gas concentration, temperature and humidity, and pressure, by decomposing the signal characteristics of each information channel, to ensure that the system can comprehensively and accurately obtain the status information of the fumigation environment and provide effective support for subsequent intelligent regulation.

4. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 3, characterized in that: Firstly, the multimodal data obtained from the multi-parameter sensor monitoring module are divided into two types, including the gas concentration of fumigation treatment and the temperature, humidity and pressure data of the fumigation environment, respectively. Two modal data sets are formed, namely the gas concentration set X and the temperature and humidity set Y. The multimodal data are fused based on the correlation between the two types of data, and an information channel is constructed to quantify the degree of information transmission between the two types of data. The specific formula is expressed as follows: Among them, P(Y|X) represents the conditional probability matrix of the first information channel, p(y i |x j ) represents the gas concentration data set X for fumigation treatment, with the value x j The temperature, humidity and pressure data set Y is y i The conditional probability of, i represents the number index of temperature, humidity and pressure data sets, j represents the number index of fumigation gas concentration data sets, n(x j ,y i ) is expressed as the gas concentration that satisfies the fumigation treatment at the same time is x j And the temperature, humidity and pressure data are y i The number of samples, n(x) represents the total number of samples with a gas concentration of x in all fumigation treatments, P(X|Y) represents the conditional probability matrix of the second information channel, p(x j |y i ) represents the temperature, humidity and pressure data set Y with the value of y i The gas concentration data set Y of the fumigation treatment is taken as y i The conditional probability of n(y) is expressed as the total number of samples with temperature, humidity and pressure data taking the value of y. Then, the marginal probability distribution P(X) and P(Y) of each sensor data are calculated, and the mutual information I(X; Y) of X and Y is quantified by the distribution. The mutual information is used to measure the amount of information sharing between sensor data. The specific formula is expressed as follows: Among them, p(x j ) and p(y i ) represents the marginal probability of X and Y, N represents the total number of multimodal data samples, n(x j ) represents the gas concentration of fumigation treatment, which is x j The total number of samples, n(y i ) represents the temperature, humidity and pressure data value is y i The total number of samples, I(X;Y) is used to quantify the amount of information shared between two data sets X and Y, p(x j ,y i ) represents the value x in the set X j And the set Y takes the value y i The joint probability of simultaneous occurrence, log, is expressed as a logarithmic function. If p(x j ,y i )>p(x j )p(y i ), it means that there is information association between the multimodal data of set X and set Y. If p(x j ,y i )≤p(x j )p(y i ), it means that the multimodal data between set X and set Y are independent of each other. Among them, I sp ec(x j ; Y) represents the data point x j The specific information contribution to the set Y, I spec (y i ; X) represents the data point y i For the specific information contribution of set X, by calculating the specific information contribution of two modal data information to the other modal data information respectively, the redundant information influence of the data point is eliminated, so as to ensure that each sensor data point can accurately reflect its environmental uniqueness in the fusion. Based on the calculated specific information amount, the value with the largest amount of information is selected in each pair of corresponding sensor data points as part of the fusion result to achieve point-by-point optimization fusion of multimodal data. The specific formula is expressed as: Among them, z k It is represented as the kth data information of the fused data set, and k is represented as the specific data point position index currently being compared and fused. By comparing specific information, it is ensured that the fused data points in the fumigation process reflect the key information characteristics of the environment to the greatest extent. Finally, all fused data points are combined into a complete data set Z, and the entropy of the fused data set is calculated to quantify the multimodal information gain and ensure the effectiveness of the fusion result. The specific formula is expressed as: Among them, H(Z) represents the entropy of the fused data set Z, z k It is represented as the kth data value after fusion, p(z k ) is represented by the data point z k The probability distribution of multimodal data is quantified by the overall information gain after fusion of multimodal data, which can ensure the effect and accuracy of the fusion algorithm. By calculating the entropy of the fused data set, the multimodal data fusion algorithm based on information channel decomposition quantifies the overall information gain after fusion of multimodal data in the fumigation environment, further ensuring the effectiveness of the algorithm.

5. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 1, characterized in that: The intelligent monitoring and regulation module of the system includes an intelligent comprehensive monitoring unit and a dynamic feedback regulation unit. The intelligent comprehensive monitoring unit proposes a multi-parameter intelligent monitoring algorithm based on a multi-layer feedforward neural network to perform deep learning and processing on the data from the sensor, ensuring that each parameter of the fumigation environment can be monitored in real time and accurately. The dynamic feedback regulation unit regulates and controls the delivery and concentration of the fumigation agent according to the analysis results of the intelligent comprehensive monitoring unit to ensure the optimized stability of the fumigation effect. Through the feedback regulation strategy, it is ensured that the system responds to the monitored changes in real time and optimizes the fumigation process.

6. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 5, characterized in that: Firstly, a multi-layer feedforward neural network model is constructed based on the data information obtained from the multimodal data processing and analysis module. The input layer, hidden layer and output layer of the neural network are defined. The input layer nodes represent the fumigation environment parameters processed by the multimodal data processing and analysis module, including multiple monitoring parameters such as gas concentration, temperature, humidity and pressure. The hidden layer nodes are used to learn the complex nonlinear relationship of the input data. The output layer nodes predict the trend of key parameters of the fumigation environment. The connection weights and bias thresholds of each layer of nodes are given initial values ​​to ensure that the neural network has a good initial state during training. The specific formula is expressed as follows: b j ＝0 Among them, w ij is the weight connecting the input node i and the hidden node j, n is the number of input layer nodes, b j It is represented as the bias of hidden node j, i is represented as the input node number index, j is represented as the hidden node number index, and U is represented as the uniform distribution function. By initializing the weights and thresholds, it provides a starting point for training for the multi-layer feedforward neural network, ensuring the randomness of the weights to avoid falling into the local optimum. Then, the input parameter data is passed to the hidden layer and the output layer in turn, and the activation value and output value of each node are calculated. The output function of the hidden layer is specifically expressed as: Among them, h j is the output value of the jth node in the hidden layer, σ is the activation function of the hidden layer, which uses a nonlinear function to capture the complex features of the input data, I is the total number of hidden nodes, and x is i The input value of the i-th node in the input layer, e, is expressed as an exponential function, and the output function of the output layer is specifically expressed as: Among them, k It represents the output value of the kth node in the output layer, reflecting the prediction result of the multi-layer feedforward neural network on the input data. k represents the index of the number of output nodes, H represents the index of the number of hidden layer nodes, and v jk It is represented as the connection weight between the jth node in the hidden layer and the kth node in the output layer, h j It is represented as the output value of the hth node in the hidden layer, which represents the value obtained after the hidden layer processes the input through the activation function, reflecting the nonlinear transformation and feature extraction capabilities of the hidden layer nodes for the input data. k It is expressed as the bias of the kth node in the output layer, which is used to adjust the activation level of the node and improve the fitting ability of the neural network. It generates the predicted value o through forward propagation calculation. k , used to evaluate the parameter change trend of the fumigation environment and provide a basis for real-time intelligent monitoring. Secondly, the error between the output value and the target value is calculated, and the weight and threshold are adjusted through the back propagation algorithm. The error propagation is divided into the gradient update process of the output layer and the hidden layer. The purpose is to minimize the prediction error. The specific formula for calculating the output layer error is expressed as: δ k =(t k -o k )o k (1-o k ) Among them, δ k Represented as the error signal of the kth neuron in the output layer, used to update the adjustment amount of the neuron and its associated weights, t k It is expressed as the target output value of the kth neuron in the output layer. The specific formula for calculating the hidden layer error is expressed as: Among them, Δ j It is represented as the error gradient of the jth neuron in the hidden layer, reflecting the contribution of the current hidden neuron to the overall error. O is represented as the total number of output nodes. The specific formula for weight update calculation is expressed as: In ij (n)=in ij (n-1)+ηδ j x i ,v j k(n)=v jk (n)+ηδ k h j Among them, η represents the learning step size, n represents the iteration number index, and finally, the forward propagation and back propagation steps are repeated until the network output error meets the set accuracy requirement, that is, the maximum number of training times is reached. The specific formula is expressed as: Where E represents the objective function for calculating the overall error in the multi-layer feedforward neural network, t k It is represented as the target value of the output node k, and ε is represented as the set error accuracy. Through multiple iterative training, the neural network gradually learns the complex relationship of multiple parameters in the fumigation environment, realizes accurate monitoring and trend prediction, and provides strong support for the intelligent monitoring of the system.

7. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 1, characterized in that: The fumigation environment control module ensures that the fumigation environment can be maintained in an ideal state by adjusting and controlling key parameters of the environment during the fumigation process, including ambient temperature, humidity and pressure parameters.

8. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 1, characterized in that: The remote maintenance and management module realizes remote access, monitoring and management of the entire system through remote communication technology, supports remote maintenance functions such as equipment status monitoring, troubleshooting, and updates and upgrades, and ensures the stability and long-term operation of the system.

9. A fumigation treatment multi-parameter comprehensive monitoring system according to claim 1, characterized in that: The system interaction and feedback module provides real-time data display, parameter adjustment function and fault warning information of the fumigation process by constructing an interactive interface, ensuring that users can intuitively understand the fumigation process and make real-time adjustments.

Images