Environmental risk monitoring system for hazardous waste storage warehouse
By designing an environmental risk monitoring system in hazardous waste storage warehouses, monitoring and analyzing temperature, humidity and gas concentration data in real time, and using machine learning algorithms to conduct risk assessment, the problem of difficulty in quantifying environmental risks in the existing technology is solved, and the risk response is achieved Safe and stable operation and risk prevention and control of waste storage process.
Patent Information
- Application Number
- CN202411476056.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-22
- Publication Date
- 2024-11-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology is difficult to quantify the environmental risks of hazardous waste storage, and lacks data analysis and risk assessment models to accurately assess potential risks and set risk levels, resulting in frequent accidents during hazardous waste storage, and major public and private property losses and personal safety hazards.
A hazardous waste storage warehouse environmental risk monitoring system was designed, including sensor module, data transmission module, data processing and storage module, risk assessment model module and early warning notification and emergency response module. By monitoring temperature, humidity and gas concentration data in real time, Use big data processing and machine learning algorithms (decision trees or neural networks) to perform risk assessment and hierarchy, and send early warning notifications and emergency response plans when risks reach thresholds.
Realize real-time environmental risk monitoring and accurate risk assessment of hazardous waste storage warehouses, timely discover potential risks, reduce the probability of accidents, ensure personnel and environmental safety, rationally allocate resources, reduce potential hazards of hazardous waste to the environment, and conform to sustainable development. Require.
Smart Images

Figure CN118982247A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of hazardous waste storage, and in particular to an environmental risk monitoring system for hazardous waste storage warehouses. Background Art
[0002] Hazardous waste refers to solid waste that is included in the national hazardous waste list or identified as having hazardous characteristics (including corrosive, toxic, flammable, reactive and infectious) according to the national hazardous waste identification standards and identification methods. In recent years, with the acceleration of urbanization and industrialization, the generation of hazardous waste has maintained a high growth rate. In addition, there are many types of hazardous wastes with complex components. The overall generation intensity is high, the disposal and utilization capacity is insufficient, and pollution accidents occur frequently, which poses a huge threat to the ecological environment and human health. The storage of hazardous wastes has potential risks, especially after the accident, the huge loss of public and private property and major personal safety hazards have aroused public concerns about the illegal storage of hazardous wastes.
[0003] Due to the potential risks in the storage of hazardous waste, the current means of environmental risk monitoring and prevention and control is to use monitoring and data collection devices. However, it is difficult to quantify environmental risks in the hazardous waste storage process. At the same time, there is a lack of data analysis and risk assessment models to accurately assess potential risks and set risk levels. In summary, in order to address these problems and strengthen environmental risk prevention and control in the hazardous waste storage process, this patent application came into being. Summary of the invention
[0004] In order to solve the technical problems existing in the background technology, the present invention proposes an environmental risk monitoring system for hazardous waste storage warehouses.
[0005] The present invention proposes an environmental risk monitoring system for hazardous waste storage warehouses, comprising: a sensor module, a data transmission module, a data processing and storage module, a risk assessment model module and an early warning notification and emergency response module, wherein the sensor module is connected to the data transmission module, the data transmission module is connected to the data processing and storage module, the data processing and storage module is connected to the risk assessment model module, and the risk assessment model module is connected to the early warning notification and emergency response module; The sensor module is used to monitor the environmental risks of hazardous waste storage warehouses in real time, collect the temperature, humidity and gas concentration in the hazardous waste storage warehouse in real time, and send the data to the data transmission gateway at the moment of collection, ensuring the timeliness and integrity of the data, providing a solid data foundation for subsequent processing and analysis Data transmission module: After receiving the data collected by the sensor module, the data transmission gateway will immediately carry out preliminary processing and encryption operations. After processing and encryption, the data transmission gateway uses the communication method of LoRa or NB-IoT network to transmit the data to the cloud server through LoRa or NB-IoT network; LoRa or NB-IoT wireless transmission technology is selected to ensure that the data collected by the sensor can be transmitted to the data processing center stably and in real time, and the data transmission gateway is configured to perform preliminary processing and encryption on the transmitted data to improve the security and reliability of data transmission. Data processing and storage module: When the cloud server successfully receives data from the data transmission gateway, it immediately starts the big data processing technology, uses the big data processing technology to clean, analyze and store the data, removes abnormal data and duplicate data during the cleaning process to ensure the accuracy and validity of the data, calculates the mean, variance and other statistical indicators of various parameters during the analysis process, and stores the processed data in the distributed database; providing strong data support for subsequent risk assessment and decision-making Build a cloud computing-based big data processing platform that can process massive amounts of real-time monitoring data and adopt a distributed storage architecture to ensure high availability and persistence of data. The data storage time is no less than 5 years. Risk assessment model module: The risk assessment model will regularly read the monitoring data stored in the distributed database at predetermined time intervals, and establish a risk assessment model by applying risk assessment algorithms and models, and using decision trees or neural network algorithms in machine learning algorithms. It will conduct a comprehensive analysis and evaluation of these data, conduct risk assessment and grade classification, and the evaluation results will be updated in real time and stored in the distributed database so that they can be queried and called at any time, providing an accurate basis for taking corresponding risk control measures in a timely manner.
[0006] Early warning notification and emergency response module: When the risk assessment result reaches the early warning threshold, the early warning notification module is immediately activated to send early warning information to relevant personnel. The notification content includes risk level, possible hazards and recommended measures. At the same time, the early warning information is recorded in the system log for subsequent query and analysis. After receiving the early warning notification, the relevant personnel will formulate a detailed emergency response plan, including personnel evacuation routes, emergency material deployment, and rescue measures, and take corresponding measures according to the emergency response plan, such as starting ventilation equipment, evacuating personnel, organizing rescue, etc. All operations and processing results in the emergency response process are fed back to the system in a timely manner to optimize and improve the subsequent risk assessment and emergency response plan.
[0007] Preferably, the sensor module comprises: Temperature sensors monitor the temperature in different areas of the warehouse with an accuracy of ±0.5°C and a measurement range of -40°C to 125°C; Humidity sensor, which measures the relative humidity in the warehouse with an accuracy of ±3%RH and a measurement range of 0%RH to 100%RH; Gas concentration sensor, measuring gas concentration in the warehouse, measuring range: 0 - 1000 ppm (parts per million), its accuracy is within the range of 0 - 200 ppm, the accuracy is ±5 ppm of the measured value, its accuracy is within the range of 201 - 1000ppm, the accuracy is ±3% of the measured value.
[0008] Preferably, in the data processing and storage module: set up: The original data set D received by the cloud server; The processed data set D' is stored in the distributed database, along with various statistical indicators; Data cleaning: For each feature column F in dataset D: Calculate the mean μ(F) and standard deviation σ(F) of the column; Eliminate the data points x that satisfy the condition |x - μ(F)| > 3σ(F) to obtain the cleaned data set D1; Remove duplicate data rows from data set D1 to obtain data set D2; Data analysis in progress: For each feature column F in dataset D2; Calculate the mean: Mean(F) = Σx / n , where x is the data value in the column and n is the number of data points; Calculate variance: Variance(F) = Σ(x - Mean(F))² / n; Data storage: The processed data D2 and the calculated mean and variance are stored in a distributed database.
[0009] Through the above algorithms, the cleaning, analysis and storage of hazardous waste storage warehouse data can be achieved.
[0010] Preferably, in the risk assessment model module, it is set as follows: The original data set D received by the cloud server; The processed data set D' is stored in the distributed database, along with various statistical indicators; Data cleaning: For each feature column F in dataset D: Calculate the mean μ(F) and standard deviation σ(F) of the column; Eliminate the data points x that satisfy the condition |x - μ(F)| > 3σ(F) to obtain the cleaned data set D1; Remove duplicate data rows from data set D1 to obtain data set D2; Data analysis in progress: For each feature column F in dataset D2; Calculate the mean: Mean(F) = Σx / n , where x is the data value in the column and n is the number of data points; Calculate variance: Variance(F) = Σ(x - Mean(F))² / n; Data storage: The processed data D2 and the calculated mean and variance are stored in a distributed database.
[0011] Preferably, in the risk assessment model module, it is set as follows: Historical monitoring data, including temperature T, humidity, various gas concentrations C1, C2, ..., Cn, and corresponding risk level labels R, which are divided into low, medium, high, and very high; Data preprocessing: The input historical monitoring data is standardized to unify the dimensions of different features: For temperature T, convert to standard value T' = (T - μ_T) / σ_T, where μ_T is the mean temperature and σ_T is the standard deviation of temperature; Do the same for humidity and gas concentration; Feature combination and construction: Construct new features, combinations of gas concentrations, and interaction terms of temperature and humidity; Model training: When using decision trees: Starting from the root node, calculate the information gain or Gini coefficient of each feature, select the feature with the largest information gain or the smallest Gini coefficient as the split feature, determine the split point, divide the data set into subsets, and repeat the above steps for each subset until the stopping condition is met; the stopping condition can be set to reach a predetermined tree depth or the number of samples in the subset is too small; When using a neural network: Initialize the weights and biases of the neural network: Forward propagation: Calculate the output value of each neuron; Backward Propagation: Adjust weights and biases based on the error between the predicted value and the actual value; Repeat forward propagation and backpropagation until the model converges or reaches the predetermined number of training rounds; Model Evaluation: Use the validation set data to input the trained model to obtain the predicted risk level R_pred, and calculate the accuracy, recall, and F1 value; Risk level classification: According to the output value of the model, set the threshold to divide the risk level: If the output value < threshold 1, the risk level is low; If threshold 1 <= output value < threshold 2, the risk level is medium; If threshold 2 <= output value < threshold 3, the risk level is high; If the output value >= threshold 3, the risk level is extremely high.
[0012] Preferably, the following is the equation algorithm for calculating precision, recall and F1 value: set up: True Positive (TP): The number of samples that are actually positive and predicted to be positive. True Negative (TN): the number of samples that are actually negative examples and are predicted to be negative examples; False Positive (FP): The number of samples that are actually negative examples but predicted to be positive examples. False Negative (FN): The number of samples that are actually positive examples but predicted to be negative examples. The accuracy is calculated as follows: Accuracy = (TP + TN) / (TP + TN + FP + FN) The calculation formula for recall is: Recall = TP / (TP + FN) The calculation formula of F1 value is: F1 = 2 * (Precision * Recall) / (Precision + Recall) Where Precision = TP / (TP + FP).
[0013] Preferably, the following is a specific equation algorithm for model training using a decision tree: set up: Assume the data set is , including features , and risk level labels ; Calculate the information gain or Gini coefficient of a feature: For features , calculate its information gain or Gini coefficient , Information Gain: , in, , Yes Category In the dataset The proportion of It is a feature The value is A subset of the data; Gini coefficient: , Select splitting features and splitting points: Select the feature with the largest information gain or the smallest Gini coefficient as the split feature, denoted as ; Determine the split point , the dataset Divided into two subsets and ; ; ; Recursively build a subtree: Subset and Repeat the steps to calculate the information gain or Gini coefficient of the feature and select the split feature and split point respectively until the stopping condition is met. The stopping condition is that the tree depth reaches the predetermined depth or the number of samples in the subset is too small. Repeat the above steps until a complete decision tree model is constructed.
[0014] Preferably, the following is a specific equation algorithm for model training using a neural network: set up: Randomly initialize the weights of the neural network and bias ; Forward propagation: For the input layer neurons, the input values are the eigenvalues constructed after standardization and combination. ; For the hidden layer neurons, the input is , the output is ,in For the Sigmoid function: ; For the output layer neurons, the input is , the output is the predicted value ; Calculation error: error ,in is the actual risk level label, is the predicted risk level value; Backward Propagation: Calculate the error gradient of the output layer:
[0015] Calculate the error gradient of the hidden layer:
[0016] Update weights: , ,in is the learning rate; Update bias: , ; Repeat the steps of forward propagation, error calculation, and back propagation until the model converges, the error is less than the predetermined value, or the predetermined number of training rounds is reached.
[0017] Preferably, the early warning notification and emergency response module, when the risk assessment result reaches the early warning threshold, sends early warning notifications to relevant personnel through SMS, email, App push and other methods. The notification content includes the risk level, possible hazards and recommended measures, formulates a detailed emergency response plan, including personnel evacuation routes, emergency material deployment, rescue measures, and takes corresponding measures according to the emergency response plan, such as starting ventilation equipment, evacuating personnel, and organizing rescue.
[0018] In the present invention, the proposed environmental risk monitoring system for hazardous waste storage warehouses has the following beneficial technical effects: 1. By real-time monitoring of key parameters such as temperature, humidity and gas concentration in hazardous waste storage warehouses, potential environmental risks can be discovered in a timely manner, and possible environmental pollution and safety accidents can be prevented. Big data and cloud computing technologies can be used for data processing and storage, abnormal and duplicate data can be eliminated, and statistical indicators such as mean and variance can be calculated to ensure data quality and timeliness, providing a reliable basis for accurate risk assessment. Risk assessment models can be established using machine learning algorithms such as decision trees or neural networks, which can comprehensively consider multiple factors for comprehensive and accurate risk assessment, and divide risk levels into different levels to make risk conditions clearer and more explicit.
[0019] 2. When the risk reaches the warning threshold, early warning notifications can be quickly sent to relevant personnel through multiple channels, and emergency response plans can be formulated at the same time, which helps to take timely measures to reduce losses and ensure the safety of personnel and the environment. This patent application helps to strengthen the environmental risk prevention and control of hazardous waste storage processes, meet the urgent needs of governments and enterprises, and alleviate the public's concerns about illegal storage of hazardous wastes. It can ensure the safe and stable operation of hazardous waste storage warehouses, reduce the probability of accidents, and protect public and private property and personal safety.
[0020] 3. It helps to rationally allocate resources to prevent and respond to risks of different levels, improve resource utilization efficiency, effectively reduce the potential harm of hazardous waste to the environment, meet the requirements of sustainable development, and promote the green development of society.
[0021] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 It is a principle block diagram of the system of the present invention; Figure 2 It is a principle block diagram of the risk assessment model module of the present invention; Figure 3 This is the selection scheme of the present invention regarding when to select a decision tree model and when to select a neural network model. DETAILED DESCRIPTION
[0023] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar symbols throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.
[0024] like Figure 1 As shown, an environmental risk monitoring system for hazardous waste storage warehouses includes: a sensor module, a data transmission module, a data processing and storage module, a risk assessment model module, and an early warning notification and emergency response module. The sensor module is connected to the data transmission module, the data transmission module is connected to the data processing and storage module, the data processing and storage module is connected to the risk assessment model module, and the risk assessment model module is connected to the early warning notification and emergency response module; The sensor module is used to monitor the environmental risks of hazardous waste storage warehouses in real time, collect the temperature, humidity and gas concentration in the hazardous waste storage warehouse in real time, and send the data to the data transmission gateway at the moment of collection, ensuring the timeliness and integrity of the data, providing a solid data foundation for subsequent processing and analysis Data transmission module: After receiving the data collected by the sensor module, the data transmission gateway will immediately carry out preliminary processing and encryption operations. After processing and encryption, the data transmission gateway uses the communication method of LoRa or NB-IoT network to transmit the data to the cloud server through LoRa or NB-IoT network; LoRa or NB-IoT wireless transmission technology is selected to ensure that the data collected by the sensor can be transmitted to the data processing center stably and in real time, and the data transmission gateway is configured to perform preliminary processing and encryption on the transmitted data to improve the security and reliability of data transmission. Data processing and storage module: When the cloud server successfully receives data from the data transmission gateway, it immediately starts the big data processing technology, uses the big data processing technology to clean, analyze and store the data, removes abnormal data and duplicate data during the cleaning process to ensure the accuracy and validity of the data, calculates the mean, variance and other statistical indicators of various parameters during the analysis process, and stores the processed data in the distributed database; providing strong data support for subsequent risk assessment and decision-making Build a cloud computing-based big data processing platform that can process massive amounts of real-time monitoring data and adopt a distributed storage architecture to ensure high availability and persistence of data. The data storage time is no less than 5 years. Risk assessment model module: The risk assessment model will regularly read the monitoring data stored in the distributed database at predetermined time intervals, and establish a risk assessment model by applying risk assessment algorithms and models, and using decision trees or neural network algorithms in machine learning algorithms. It will conduct a comprehensive analysis and evaluation of these data, conduct risk assessment and grade classification, and the evaluation results will be updated in real time and stored in the distributed database so that they can be queried and called at any time, providing an accurate basis for taking corresponding risk control measures in a timely manner.
[0025] Early warning notification and emergency response module: When the risk assessment result reaches the early warning threshold, the early warning notification module is immediately activated to send early warning information to relevant personnel. The notification content includes risk level, possible hazards and recommended measures. At the same time, the early warning information is recorded in the system log for subsequent query and analysis. After receiving the early warning notification, the relevant personnel will formulate a detailed emergency response plan, including personnel evacuation routes, emergency material deployment, and rescue measures, and take corresponding measures according to the emergency response plan, such as starting ventilation equipment, evacuating personnel, organizing rescue, etc. All operations and processing results in the emergency response process are fed back to the system in a timely manner to optimize and improve the subsequent risk assessment and emergency response plan.
[0026] In an optional embodiment, the sensor module includes: Temperature sensors monitor the temperature in different areas of the warehouse with an accuracy of ±0.5°C and a measurement range of -40°C to 125°C; Humidity sensor, which measures the relative humidity in the warehouse with an accuracy of ±3%RH and a measurement range of 0%RH to 100%RH; Gas concentration sensor, measuring gas concentration in the warehouse, measuring range: 0 - 1000 ppm (parts per million), its accuracy is within the range of 0 - 200 ppm, the accuracy is ±5 ppm of the measured value, its accuracy is within the range of 201 - 1000ppm, the accuracy is ±3% of the measured value.
[0027] In an optional embodiment, in the data processing and storage module: set up: The original data set D received by the cloud server; The processed data set D' is stored in the distributed database, along with various statistical indicators; Data cleaning: For each feature column F in dataset D: Calculate the mean μ(F) and standard deviation σ(F) of the column; Eliminate the data points x that satisfy the condition |x - μ(F)| > 3σ(F) to obtain the cleaned data set D1; Remove duplicate data rows from data set D1 to obtain data set D2; Data analysis in progress: For each feature column F in dataset D2; Calculate the mean: Mean(F) = Σx / n , where x is the data value in the column and n is the number of data points; Calculate variance: Variance(F) = Σ(x - Mean(F))² / n; Data storage: The processed data D2 and the calculated mean and variance are stored in a distributed database.
[0028] Through the above algorithms, the cleaning, analysis and storage of hazardous waste storage warehouse data can be achieved.
[0029] like Figure 2 As shown: In an optional embodiment, in the risk assessment model module, it is set that: The original data set D received by the cloud server; The processed data set D' is stored in the distributed database, along with various statistical indicators; Data cleaning: For each feature column F in dataset D: Calculate the mean μ(F) and standard deviation σ(F) of the column; Eliminate the data points x that satisfy the condition |x - μ(F)| > 3σ(F) to obtain the cleaned data set D1; Remove duplicate data rows from data set D1 to obtain data set D2; Data analysis in progress: For each feature column F in dataset D2; Calculate the mean: Mean(F) = Σx / n , where x is the data value in the column and n is the number of data points; Calculate variance: Variance(F) = Σ(x - Mean(F))² / n; Data storage: The processed data D2 and the calculated mean and variance are stored in a distributed database.
[0030] Preferably, in the risk assessment model module, it is set as follows: Historical monitoring data, including temperature T, humidity, various gas concentrations C1, C2, ..., Cn, and corresponding risk level labels R, which are divided into low, medium, high, and very high; Data preprocessing: The input historical monitoring data is standardized to unify the dimensions of different features: For temperature T, convert to standard value T' = (T - μ_T) / σ_T, where μ_T is the mean temperature and σ_T is the standard deviation of temperature; Do the same for humidity and gas concentration; Feature combination and construction: Construct new features, combinations of gas concentrations, and interaction terms of temperature and humidity; Model training: When using decision trees: Starting from the root node, calculate the information gain or Gini coefficient of each feature, select the feature with the largest information gain or the smallest Gini coefficient as the split feature, determine the split point, divide the data set into subsets, and repeat the above steps for each subset until the stopping condition is met; the stopping condition can be set to reach a predetermined tree depth or the number of samples in the subset is too small; When using a neural network: Initialize the weights and biases of the neural network: Forward propagation: Calculate the output value of each neuron; Backward Propagation: Adjust weights and biases based on the error between the predicted value and the actual value; Repeat forward propagation and backpropagation until the model converges or reaches the predetermined number of training rounds; Model Evaluation: Use the validation set data to input the trained model to obtain the predicted risk level R_pred, and calculate the accuracy, recall, and F1 value; Risk level classification: According to the output value of the model, set the threshold to divide the risk level: If the output value < threshold 1, the risk level is low; If threshold 1 <= output value < threshold 2, the risk level is medium; If threshold 2 <= output value < threshold 3, the risk level is high; If the output value >= threshold 3, the risk level is extremely high.
[0031] Through the above algorithm, an environmental risk assessment model for hazardous waste storage warehouses is established, which can comprehensively consider multiple factors and realize the division of risk levels.
[0032] In an optional embodiment, the following is an equation algorithm for calculating precision, recall, and F1 value: set up: True Positive (TP): The number of samples that are actually positive and predicted to be positive. True Negative (TN): the number of samples that are actually negative examples and are predicted to be negative examples; False Positive (FP): The number of samples that are actually negative examples but predicted to be positive examples. False Negative (FN): The number of samples that are actually positive examples but predicted to be negative examples. The accuracy is calculated as follows: Accuracy = (TP + TN) / (TP + TN + FP + FN) The calculation formula for recall is: Recall = TP / (TP + FN) The calculation formula of F1 value is: F1 = 2 * (Precision * Recall) / (Precision + Recall) Where Precision = TP / (TP + FP).
[0033] In an optional embodiment, the following is a specific equation algorithm for model training using a decision tree: set up: Assume the data set is , including features , and risk level labels ; Calculate the information gain or Gini coefficient of a feature: For features , calculate its information gain or Gini coefficient , Information Gain: , in,
[0034] Yes Category In the dataset The proportion of It is a feature The value is A subset of the data; Gini coefficient:
[0035] Select splitting features and splitting points: Select the feature with the largest information gain or the smallest Gini coefficient as the split feature, denoted as ; Determine the split point , the dataset Divided into two subsets and ; ; ; Recursively build a subtree: Subset and Repeat the steps to calculate the information gain or Gini coefficient of the feature and select the split feature and split point respectively until the stopping condition is met. The stopping condition is that the tree depth reaches the predetermined depth or the number of samples in the subset is too small. Repeat the above steps until a complete decision tree model is constructed.
[0036] In an optional embodiment, the following is a specific equation algorithm for model training using a neural network: set up: Randomly initialize the weights of the neural network and bias ; Forward propagation: For the input layer neurons, the input values are the eigenvalues constructed after standardization and combination. ; For the hidden layer neurons, the input is , the output is ,in For the Sigmoid function: ; For the output layer neurons, the input is , the output is the predicted value ; Calculation error: error ,in is the actual risk level label, is the predicted risk level value; Backward Propagation: Calculate the error gradient of the output layer:
[0037] Calculate the error gradient of the hidden layer:
[0038] Update weights: , ,in is the learning rate; Update bias: , ; Repeat the steps of forward propagation, error calculation, and back propagation until the model converges, the error is less than the predetermined value, or the predetermined number of training rounds is reached.
[0039] In an optional embodiment, the early warning notification and emergency response module, when the risk assessment result reaches the early warning threshold, sends an early warning notification to relevant personnel through various methods such as SMS, email, App push, etc. The notification content includes the risk level, possible hazards and recommended measures, and formulates a detailed emergency response plan, including personnel evacuation routes, emergency material deployment, and rescue measures. Appropriate measures are taken in accordance with the emergency response plan, such as starting ventilation equipment, evacuating personnel, and organizing rescue.
[0040] like Figure 3 As shown: In an optional embodiment, the following is a selection scheme on when to choose a decision tree model and when to choose a neural network model: If the relationship between data features is relatively linear and the amount of data is relatively small, use a decision tree for model training; If there are complex nonlinear relationships between data features and there is enough data to support training, use a neural network for model training; If you need a clear and intuitive explanation of the model's decision-making process, use a decision tree to train the model; If the interpretability requirement is not high and the prediction performance of the model is more important, use a neural network for model training; If computing resources are limited and training time is short, using decision trees for model training can usually train and get results faster; If there are sufficient computing resources and time for long-term training, using neural networks for model training may be able to tap into deeper feature representations.
[0041] In this application: By real-time monitoring of key parameters such as temperature, humidity and gas concentration in hazardous waste storage warehouses, potential environmental risks can be discovered in a timely manner, and possible environmental pollution and safety accidents can be prevented. Big data and cloud computing technologies can be used for data processing and storage, abnormal and duplicate data can be eliminated, and statistical indicators such as mean and variance can be calculated to ensure data quality and timeliness, providing a reliable foundation for accurate risk assessment. Risk assessment models can be established using machine learning algorithms such as decision trees or neural networks, which can comprehensively consider multiple factors for comprehensive and accurate risk assessments, and divide risk levels into different levels to make risk conditions clearer and more specific.
[0042] When the risk reaches the warning threshold, early warning notifications can be quickly sent to relevant personnel through multiple channels, and emergency response plans can be formulated at the same time, which helps to take timely measures to reduce losses and ensure the safety of personnel and the environment. This patent application helps to strengthen the environmental risk prevention and control of hazardous waste storage processes, meet the urgent needs of governments and enterprises, and alleviate the public's concerns about illegal storage of hazardous wastes. It can ensure the safe and stable operation of hazardous waste storage warehouses, reduce the probability of accidents, and protect public and private property and personal safety.
[0043] It helps to rationally allocate resources to prevent and respond to risks of different levels, improve resource utilization efficiency, effectively reduce the potential harm of hazardous waste to the environment, meet the requirements of sustainable development, and promote the green development of society.
[0044] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.
[0045] In the embodiments provided by the present invention, it should be understood that the disclosed system can be implemented in other ways. For example, the above-described embodiments of the invention are only illustrative, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.
[0046] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0047] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of hardware plus software functional modules.
[0048] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the basic features of the present invention.
[0049] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. An environmental risk monitoring system for hazardous waste storage warehouses, characterized in that: include: A sensor module, a data transmission module, a data processing and storage module, a risk assessment model module, and an early warning notification and emergency response module. The sensor module is connected to the data transmission module, the data transmission module is connected to the data processing and storage module, the data processing and storage module is connected to the risk assessment model module, and the risk assessment model module is connected to the early warning notification and emergency response module; Sensor module: used to monitor the environmental risks of hazardous waste storage warehouses in real time, collect temperature, humidity and gas concentration in hazardous waste storage warehouses in real time, and send the data to the data transmission gateway at the moment of collection; Data transmission module: After receiving the data collected by the sensor module, the data transmission gateway will immediately carry out preliminary processing and encryption operations. After the processing and encryption are completed, the data transmission gateway uses the communication method of the LoRa or NB-IoT network to transmit the data to the cloud server through the LoRa or NB-IoT network; Data processing and storage module: When the cloud server successfully receives data from the data transmission gateway, it immediately starts the big data processing technology to clean, analyze and store the data. During the cleaning process, abnormal data and duplicate data are removed. During the analysis process, statistical indicators such as the mean and variance of various parameters are calculated, and the processed data is stored in the distributed database. Risk assessment model module: The risk assessment model will regularly read the monitoring data stored in the distributed database at predetermined time intervals, and establish a risk assessment model by applying risk assessment algorithms and models, and using decision trees or neural network algorithms in machine learning algorithms to conduct a comprehensive analysis and assessment of these data, conduct risk assessment and grade classification, and the assessment results will be updated in real time and stored in the distributed database; Early warning notification and emergency response module: When the risk assessment results reach the early warning threshold, the early warning notification module will be immediately activated to send early warning information to relevant personnel. At the same time, the early warning information will be recorded in the system log. After receiving the early warning notification, the relevant personnel will formulate a detailed emergency response plan and take corresponding measures in accordance with the emergency response plan. All operations and processing results in the emergency response process will be promptly fed back to the system.
2. The environmental risk monitoring system for hazardous waste storage warehouses according to claim 1 is characterized in that: The sensor module includes: Temperature sensors monitor the temperature in different areas of the warehouse with an accuracy of ±0.5°C and a measurement range of -40°C to 125°C; Humidity sensor, which measures the relative humidity in the warehouse with an accuracy of ±3%RH and a measurement range of 0%RH to 100%RH; Gas concentration sensor, measuring gas concentration in the warehouse, measuring range: 0 - 1000 ppm, its accuracy is within the range of 0 -200 ppm, the accuracy is ±5 ppm of the measured value, its accuracy is within the range of 201 - 1000 ppm, the accuracy is ±3% of the measured value.
3. The environmental risk monitoring system for hazardous waste storage warehouses according to claim 1 is characterized in that: In the data processing and storage module: set up: The original data set D received by the cloud server; The processed data set D' is stored in the distributed database, along with various statistical indicators; Data cleaning: For each feature column F in dataset D: Calculate the mean μ(F) and standard deviation σ(F) of the column; Eliminate the data points x that satisfy the condition |x - μ(F)| > 3σ(F) to obtain the cleaned data set D1; Remove duplicate data rows from data set D1 to obtain data set D2; Data analysis in progress: For each feature column F in dataset D2; Calculate the mean: Mean(F) = Σx / n , where x is the data value in the column and n is the number of data points; Calculate variance: Variance(F) = Σ(x - Mean(F))² / n; Data storage: The processed data D2 and the calculated mean and variance are stored in a distributed database.
4. The environmental risk monitoring system for hazardous waste storage warehouses according to claim 1 is characterized in that: In the risk assessment model module, set: Historical monitoring data, including temperature T, humidity, various gas concentrations C1, C2, ..., Cn, and corresponding risk level labels R, which are divided into low, medium, high, and very high; Data preprocessing: The input historical monitoring data is standardized to unify the dimensions of different features: For temperature T, convert to standard value T' = (T - μ_T) / σ_T, where μ_T is the mean temperature and σ_T is the standard deviation of temperature; Do the same for humidity and gas concentration; Feature combination and construction: Construct new features, combinations of gas concentrations, and interaction terms of temperature and humidity; Model training: When using decision trees: Starting from the root node, calculate the information gain or Gini coefficient of each feature, select the feature with the largest information gain or the smallest Gini coefficient as the split feature, determine the split point, divide the data set into subsets, and repeat the steps of starting from the root node, calculating the information gain or Gini coefficient of each feature, selecting the feature with the largest information gain or the smallest Gini coefficient as the split feature, and determining the split point for each subset until the stopping condition is met; When using a neural network: Initialize the weights and biases of the neural network: Forward propagation: Calculate the output value of each neuron; Backward Propagation: Adjust weights and biases based on the error between the predicted value and the actual value; Repeat forward propagation and backpropagation until the model converges or reaches the predetermined number of training rounds; Model Evaluation: Use the validation set data to input the trained model to obtain the predicted risk level R_pred, and calculate the accuracy, recall, and F1 value; Risk level classification: According to the output value of the model, set the threshold to divide the risk level: If the output value < threshold 1, the risk level is low; If threshold 1 <= output value < threshold 2, the risk level is medium; If threshold 2 <= output value < threshold 3, the risk level is high; If the output value >= threshold 3, the risk level is extremely high.
5. The environmental risk monitoring system for hazardous waste storage warehouses according to claim 4 is characterized in that: Accuracy refers to the proportion of correctly classified samples to the total number of samples; Recall refers to the ratio of the number of samples correctly predicted as positive examples to the number of actual positive examples. F1-score is an evaluation indicator that comprehensively considers accuracy and recall. It is the harmonic mean of accuracy and recall. The following is the equation algorithm for calculating precision, recall and F1 value: set up: True positive TP: the number of samples that are actually positive and predicted to be positive; True negative examples TN: the number of samples that are actually negative examples and predicted to be negative examples; False positive examples FP: the number of samples that are actually negative examples but predicted to be positive examples; False negative examples FN: the number of samples that are actually positive examples but predicted to be negative examples; The accuracy is calculated as follows: Accuracy = (TP + TN) / (TP + TN + FP + FN) The calculation formula for recall is: Recall = TP / (TP + FN) The calculation formula of F1 value is: F1 = 2 * (Precision * Recall) / (Precision + Recall) Where, Precision = TP / (TP + FP).
6. The environmental risk monitoring system for hazardous waste storage warehouses according to claim 4 is characterized in that: The following is the specific equation algorithm for model training using decision trees: set up: Assume the data set is , including features , and risk level labels ; Calculate the information gain or Gini coefficient of a feature: For features , calculate its information gain or Gini coefficient , Information Gain: , in, , Yes Category In the dataset The proportion of It is a feature The value is A subset of the data; Gini coefficient: , Select splitting features and splitting points: Select the feature with the largest information gain or the smallest Gini coefficient as the split feature, denoted as ; Determine the split point , the data set Divided into two subsets and ; ; ; Recursively build a subtree: Subset and Repeat the steps to calculate the information gain or Gini coefficient of the feature and select the split feature and split point respectively until the stopping condition is met. The stopping condition is that the tree depth reaches the predetermined depth or the number of samples in the subset is too small. Repeat the above steps until a complete decision tree model is constructed.
7. The environmental risk monitoring system for hazardous waste storage warehouses according to claim 4 is characterized in that: The following is the specific equation algorithm for model training using neural network: set up: Randomly initialize the weights of the neural network and bias ; Forward propagation: For the input layer neurons, the input values are the eigenvalues constructed after standardization and combination. ; For the hidden layer neurons, the input is , the output is ,in is the Sigmoid function: ; For the output layer neurons, the input is , the output is the predicted value ; Calculation error: error ,in is the actual risk level label, is the predicted risk level value; Backward Propagation: Calculate the error gradient of the output layer: , Calculate the error gradient of the hidden layer: , Update weights: , ,in is the learning rate; Update bias: , ; Repeat the steps of forward propagation, error calculation, and back propagation until the model converges, the error is less than the predetermined value, or the predetermined number of training rounds is reached.
8. The environmental risk monitoring system for hazardous waste storage warehouses according to claim 1 is characterized in that: The early warning notification and emergency response module sends early warning notifications to relevant personnel through SMS, email, App push, etc. when the risk assessment results reach the early warning threshold. The notification content includes risk level, possible hazards and recommended measures. A detailed emergency response plan is formulated, including personnel evacuation routes, emergency material allocation, and rescue measures. Appropriate measures are taken in accordance with the emergency response plan, such as starting ventilation equipment, evacuating personnel, and organizing rescue.
Citation Information
Patent Citations
Intelligent risk evaluation system for enterprise safety production
CN117455245A
Hazardous chemical substance storage early warning method and system based on improved artificial neural network
CN117455365A
Intelligent hazardous chemical substance risk analysis and evaluation system
CN118095854A
Risk Assessment And Prioritization Framework
US20120215575A1
Tailing pond risk monitoring and early-warning system based on internet of things
WO2023061039A1
Cited By
Dangerous cargo risk early warning method and system and computer program
CN119359057A
Factory dangerous chemical supervision system
CN119439859A
Hazardous waste informatization risk assessment and supervision system based on AI artificial intelligence
CN119578899A
Hazardous article warehouse risk management method and device based on sensing equipment
CN120181751A
Intelligent stereoscopic warehouse monitoring system
CN120942793A
