A mine environment safety early warning method based on BP neural network
Through the BP neural network, an intelligent early warning model was established, which solved the problem that traditional early warning methods could not analyze multiple environmental impact indicators, achieved high-accurate environmental safety warning, and prevented the occurrence of safety accidents.
Patent Information
- Application Number
- CN202210099817.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-27
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2042-01-27
AI Technical Summary
The traditional mining area environmental safety warning method is based on threshold warning, and cannot effectively analyze the comprehensive impact of multiple environmental impact indicators on environmental safety, resulting in inaccurate early warning.
Using a BP neural network-based method, the mining area environmental data is collected for preprocessing and annotation, a BP neural network model is established, and the mining area environmental data is mined by the BP neural network to form an intelligent early warning model to achieve a comprehensive early warning of environmental safety.
It improves the accuracy of the mining area environmental safety warning model, can promptly detect and warn of safety hazards, effectively prevent accidents, and has high information extraction capabilities.
Smart Images

Figure CN114417729B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of artificial intelligence, and in particular to a mining area environmental safety early warning method based on BP neural network. Background Art
[0002] Early warning of mining area environmental safety is the core issue of mining area environmental safety. Traditional early warning methods are mainly based on threshold warning methods, but this early warning method has inherent shortcomings. The threshold-based early warning method has the technical problem of prompting an alarm when a certain indicator exceeds the threshold, and cannot effectively analyze the impact of multiple environmental impact indicators on environmental safety. Relying solely on threshold alarms has the problem of serious inaccurate early warnings. Summary of the invention
[0003] The purpose of the present invention is to provide a mining area environmental safety early warning method based on BP neural network, which is easy to operate, can provide comprehensive early warning for environmental safety, avoid environmental safety hazards, and solve the problems raised in the above background technology.
[0004] To achieve the above object, the present invention provides the following technical solutions:
[0005] A mining area environmental safety early warning method based on BP neural network includes the following steps:
[0006] S1. Collect mining area environmental data using a data collection module of a mining area environmental monitoring platform;
[0007] S2. Preprocess the collected mining area environmental data, fill in missing data and remove redundant data, and manually label them into three categories: normal, warning and alarm;
[0008] The preprocessed data is normalized to form a normalized data set (x, y), and the normalized data set is divided into a training set and a test set; wherein x represents the mining area environmental data, and y represents whether the mining area environmental data is normal, warning or alarm;
[0009] S3. According to the characteristics of the mining area environmental data, a model framework that meets the characteristics is determined, and a BP neural network is established; the BP neural network includes an input layer, a hidden layer and an output layer;
[0010] The establishment of BP neural network includes the following steps:
[0011] S301, network initialization: determine the number of input layer nodes n, the number of hidden layer nodes p, and the number of output layer nodes m of the BP neural network according to the input and output sequence (x, y), initialize the connection weights ω between the input layer, hidden layer, and output layer neurons, initialize the hidden layer bias a, the output layer bias b, and give the learning rate and neuron activation function;
[0012] S302, hidden layer output calculation: according to the input vector X, ω and a, calculate the output H of the hidden layer;
[0013] Output layer output calculation: According to the output H of the hidden layer, the connection weight ω and the bias b, the output O of the BP neural network is calculated;
[0014] S303, error calculation: according to the BP neural network prediction output O and the expected output Y, calculate the BP neural network prediction error E;
[0015] S304, weight and bias update: update the network connection weight and bias according to the network prediction error E;
[0016] Determine whether the algorithm iteration is finished, if not, return to step S302;
[0017] S4, using the training set in step S2 to train the BP neural network in step S3 to form a mining area environment early warning model; using the mining area environment data in the training set as the data of the BP neural network, and outputting the prediction result of the BP neural network, and comparing the prediction result with the manual label relative to the mining area environment data in the input training set; adjusting the internal parameters of the BP neural network according to the deviation value between the prediction result and the manual label, so as to form an accurate mining area environment data early warning model according to the mining area environment data;
[0018] S5, judging whether there is a problem with the mining area environment early warning model in step S3 according to the training set in step S2 and the result of the performance test; if there is a problem, repeating steps S3 to S5 until the mining area environment early warning model passes the performance test;
[0019] S6, using the mining area environment early warning model that has passed the performance test in step S5 in actual mining area environment detection, using the data acquisition module to collect real-time data of the mining area, and applying it as input to the mining area environment early warning model, the mining area environment early warning model outputs the manual annotation label corresponding to the real-time data;
[0020] S7. When the artificial label output in step S6 is real-time data belonging to early warning data or alarm data, the alarm function is triggered; otherwise, the alarm function is not triggered.
[0021] Furthermore, the process of the signal forward propagation is: the signal passes through the hidden layer from the input layer and finally reaches the output layer; the process of establishing the hidden layer is:
[0022]
[0023] Among them, H j represents the output result of the jth hidden layer, g represents the activation function, i represents the number of input nodes, ω ij represents the weight from the input layer to the hidden layer, x i represents the input of the i-th neuron in the input layer, a j Represents the bias from the input layer to the hidden layer;
[0024] The process of establishing the output layer is as follows:
[0025]
[0026] Among them, O k represents the output of the kth neuron in the output layer, ω jk represents the weight from the hidden layer to the output layer, b k Represents the bias from the hidden layer to the output layer.
[0027] Furthermore, the error propagation adopts the gradient descent method, and the calculation formula of the loss function is:
[0028]
[0029] Where, E represents the error between the predicted value and the true value, m represents the number of output layer nodes, and Y k Indicates the expected output.
[0030] Furthermore, the correction amount of the weight and bias is proportional to the gradient of the loss function to the node, and the process of establishing the corrected weight and bias is:
[0031]
[0032] Q ij =ω ij +Δω j ,Y ij =b j +Δb j .
[0033] Compared with the prior art, the present invention has the following beneficial effects:
[0034] The present invention provides a mine environment safety early warning method based on BP neural network, establishes a mine environment safety intelligent early warning model according to the collected mine environment data, and mines the mine environment data through BP neural network, so as to improve the accuracy of the mine environment safety early warning model in diagnosing safety hazards. The present invention establishes a mine environment safety intelligent early warning model based on mine environment data, uses the mine environment data collected in real time to directly and accurately judge the safety hazards with errors through the mine environment safety early warning model, and timely warns to prevent accidents; uses BP neural network to deeply mine hidden safety hazard information, has high information extraction ability, and can bring better results for mine environment safety intelligent early warning; the conclusions obtained are accurate, and safety hazards can be discovered and warned in time, effectively preventing the occurrence of safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION
[0036] 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.
[0037] Example 1
[0038] See also Figure 1 This embodiment provides a mining area environmental safety early warning method based on BP neural network, collects mining area environmental data, performs data preprocessing, obtains a normalized sample set, and then designs a reasonable division method to divide the samples in the normalized sample set to form a training set and a test set of samples; the samples in the training set are used to train the BP neural network to build a mining area environmental safety early warning model; the internal weight parameters of the early warning model are adjusted according to the deviation between the prediction results of the mining area environmental safety early warning model and the manually labeled labels corresponding to the measured mining area environmental data; the samples in the test set are used to perform performance testing on the mining area environmental safety early warning model, thereby improving the accuracy of the diagnosis of the mining area environmental safety early warning model, which specifically includes the following steps:
[0039] S1. Collect mining area environmental data using a data acquisition module of a mining area environmental monitoring platform; the mining area environmental data includes chemical oxygen demand (COD), hydrogen ion concentration index (PH), particulate matter with an aerodynamic equivalent diameter of less than or equal to 10 microns (PM10) in ambient air, noise, flow, temperature and humidity;
[0040] S2, manually labeling the collected mining area environmental data to determine whether each set of mining area environmental data belongs to early warning data;
[0041] The mining area environmental data is subjected to data preprocessing operations, including missing value filling, normalization processing, etc., to form a normalized sample set (x, y), and the normalized data set is divided into a training set data_train(x, y) and a test set test_train(x, y); wherein x represents the mining area environmental data, and y represents whether the mining area environmental data is normal, warning, or alarm; wherein the training set data_train(x, y) performs performance training on the BP neural network model to form a mining area environmental safety warning model, and the test set test_train(x, y) performs performance testing on the mining area environmental safety warning model to determine whether the mining area environmental safety warning model can pass the performance test;
[0042] S3. According to the characteristics of the mining area environmental data, determine a model framework that meets the characteristics and establish a BP neural network; the BP neural network includes an input layer, a convolution layer and an output layer;
[0043] The establishment of the BP neural network comprises the following steps:
[0044] S301, network initialization: according to the input and output sequence, determine the number of input layer nodes of the BP neural network as n, the number of hidden layer nodes as l, the number of output layer nodes as m, and the weight ω from the input layer to the hidden layer ij , the weight from the hidden layer to the output layer ω jk , the bias from the input layer to the hidden layer a j , the bias from the hidden layer to the output layer is b k , the learning rate is η and the activation function is g(x); the activation function from the input layer to the hidden layer is the tan function, and the activation function from the hidden layer to the output layer is the sigmoid function. The calculation process of the Sigmoid function is:
[0045]
[0046] The output range of the Sigmoid function is 0 to 1. Since the output value is limited to 0 to 1, it normalizes the output of each neuron;
[0047] The calculation process of the tanh activation function is:
[0048]
[0049] The output range of the tanh function is the interval [-1,1];
[0050] S302, hidden layer output calculation: According to the input vectors X, w and a, the output H of the hidden layer is calculated. The calculation process of the hidden layer output is:
[0051]
[0052] Among them, H j represents the output result of the jth node in the hidden layer, g represents the activation function, i represents the number of input nodes, ω ij represents the weight from the input layer to the hidden layer, x i represents the input of the i-th neuron in the input layer, a j Represents the bias from the input layer to the hidden layer;
[0053] Output layer output calculation: According to the output H of the hidden layer, the connection weight w and the bias b are used to calculate the output O of the BP neural network. The calculation process of the output layer is:
[0054]
[0055] O k represents the output result of the kth neuron, g is the activation function of the neuron, the sigmoid activation function is used as the output layer activation function, and l represents the number of nodes in the output layer, which represents how many categories the project has to be classified;
[0056] S303, error calculation: Calculate the BP neural network prediction error E based on the BP neural network prediction output O and the expected output Y. The error calculation process is:
[0057]
[0058] S304, weight and bias update: update the network connection weight and bias according to the network prediction error E, and the calculation process of weight update is:
[0059]
[0060] The calculation process of bias update is:
[0061]
[0062] Determine whether the algorithm iteration is finished, if not, return to step S302;
[0063] S4, using the training set in step S2 to train the BP neural network in step S3 to form a mining area environmental safety early warning model; using the mining area environmental data in the training set as the input of the BP neural network, and outputting the prediction result of the BP neural network, and comparing the prediction result with the manually annotated label corresponding to the mining area environmental data in the training set to which the input belongs; adjusting the internal parameters of the BP neural network model according to the deviation value between the prediction result and the manually annotated label, so as to form a mining area environmental safety early warning model that can provide accurate early warning based on the mining area environmental data;
[0064] S5, testing the mining area environmental safety early warning model in step S4 according to the test set in step S2, and judging whether there is a problem with the mining area environmental safety early warning model according to the result of the performance test; if there is a problem, repeating steps S3-S5 until the mining area environmental safety early warning model passes the performance test;
[0065] S6, applying the mining area environmental safety early warning model that has passed the performance test in step S5 to the actual mining area environmental data, using the data acquisition module to collect real-time data of the mining area, and applying it as input to the mining area environmental safety early warning model, the mining area environmental safety early warning model outputs the manual annotation labels corresponding to the real-time data;
[0066] S7. When the manually annotated label output in step S6 is real-time data belonging to early warning or alarm data, the alarm function is triggered, otherwise, the alarm function is not triggered.
[0067] This embodiment uses 80% of the normalized sample set as a training set and 20% as a test set. It uses the method proposed in the present invention and the artificial intelligence algorithm model designed to issue safety warnings for abnormalities in the mining area environmental monitoring data, determine whether there are safety hazards, and output the abnormal conditions of the mining area environment, which can make the success rate of mining area environmental safety warnings reach 98.68%.
[0068] In summary: the present invention provides a mine environmental safety early warning method based on BP neural network, which establishes a mine environmental safety intelligent early warning model based on the collected mine environmental data, and mines the mine environmental data through the BP neural network, thereby improving the accuracy of the mine environmental safety early warning model in diagnosing safety hazards. The present invention establishes a mine environmental safety intelligent early warning model based on mine environmental data, and uses the mine environmental data collected in real time to directly and accurately judge the safety hazards through the mine environmental safety early warning model, and timely warns to prevent accidents; uses the BP neural network to deeply mine the hidden safety hazard information, has a high information extraction ability, and can bring better results for the mine environmental safety intelligent early warning; the conclusions obtained are accurate, and safety hazards can be discovered and warned in time, effectively preventing the occurrence of safety accidents.
[0069] 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. A mining area environmental safety early warning method based on BP neural network, characterized in that: The following steps are involved: S1. Collect mining area environmental data using a data collection module of a mining area environmental monitoring platform; S2. Preprocess the collected mining area environmental data, fill in missing data and remove redundant data, and manually label them into three categories: normal, warning and alarm; The preprocessed data is normalized to form a normalized data set (x, y), and the normalized data set is divided into a training set and a test set; wherein x represents the mining area environmental data, and y represents whether the mining area environmental data is normal, warning or alarm; S3. According to the characteristics of the mining area environmental data, a model framework that meets the characteristics is determined, and a BP neural network is established; the BP neural network includes an input layer, a hidden layer and an output layer; The establishment of BP neural network includes the following steps: S301, network initialization: determine the number of input layer nodes n, hidden layer nodes p, and output layer nodes m of the BP neural network according to the input and output sequence (x, y), initialize the connection weights ω between the input layer, hidden layer, and output layer neurons, initialize the hidden layer bias a, the output layer bias b, and give the learning rate and neuron activation function; S302, hidden layer output calculation: according to the input vector X, ω and a, calculate the output H of the hidden layer; Output layer output calculation: According to the output H of the hidden layer, the connection weight ω and the bias b, the output O of the BP neural network is calculated; S303, error calculation: according to the BP neural network prediction output O and the expected output Y, calculate the BP neural network prediction error E; S304, weight and bias update: update the network connection weight and bias according to the network prediction error E; Determine whether the algorithm iteration is finished, if not, return to step S302; S4, using the training set in step S2 to train the BP neural network in step S3 to form a mining area environment early warning model; using the mining area environment data in the training set as the data of the BP neural network, and outputting the prediction result of the BP neural network, and comparing the prediction result with the manual label relative to the mining area environment data in the input training set; adjusting the internal parameters of the BP neural network according to the deviation value between the prediction result and the manual label, so as to form an accurate mining area environment data early warning model according to the mining area environment data; S5, judging whether there is a problem with the mining area environment early warning model in step S3 according to the training set in step S2 and the result of the performance test; if there is a problem, repeating steps S3 to S5 until the mining area environment early warning model passes the performance test; S6, using the mining area environment early warning model that has passed the performance test in step S5 in actual mining area environment detection, using the data acquisition module to collect real-time data of the mining area, and applying it as input to the mining area environment early warning model, the mining area environment early warning model outputs the manual annotation label corresponding to the real-time data; S7. When the artificial label output in step S6 is real-time data belonging to early warning data or alarm data, the alarm function is triggered; otherwise, the alarm function is not triggered.
2. A mining area environmental safety early warning method based on BP neural network as claimed in claim 1, characterized in that: The propagation process of the BP neural network in S3 is the process of signal forward propagation: the signal passes through the hidden layer from the input layer and finally reaches the output layer; the establishment process of the hidden layer is: Among them, H j represents the output result of the jth hidden layer, g represents the activation function, i represents the number of input nodes, ω ij represents the weight from the input layer to the hidden layer, x i represents the input of the i-th neuron in the input layer, a j Represents the bias from the input layer to the hidden layer; The process of establishing the output layer is as follows: Among them, O k represents the output of the kth neuron in the output layer, ω jk represents the weight from the hidden layer to the output layer, b k Represents the bias from the hidden layer to the output layer.
3. A mining area environmental safety early warning method based on BP neural network as claimed in claim 2, characterized in that: The error propagation adopts the gradient descent method, and the calculation formula of the loss function is: Among them, E represents the error between the predicted value and the true value, m represents the number of nodes in the output layer, and Y k Indicates the expected output.
4. A mining area environmental safety early warning method based on BP neural network as claimed in claim 3, characterized in that: The correction amount of the weight and bias is proportional to the gradient of the loss function to the node. The process of establishing the corrected weight and bias is: Q ij =ω ij +See j ,Y ij =b j +Δb j 。
Citation Information
Patent Citations
On-line calculation method of probabilistic power flow based on BP neural network
CN109117951A
Fault diagnosis method of spacecraft based on artificial intelligence
CN109255441A
Method for reducing fire false alarm rate based on GA-BP neural network algorithm
CN111783940A
Central air conditioner cooling load prediction method based on BP neural network
WO2015172560A1
