CN117454233A - Safety production management method and system based on positioning identification - Google Patents
Safety production management method and system based on positioning identification Download PDFInfo
- Publication number
- CN117454233A CN117454233A CN202311782147.6A CN202311782147A CN117454233A CN 117454233 A CN117454233 A CN 117454233A CN 202311782147 A CN202311782147 A CN 202311782147A CN 117454233 A CN117454233 A CN 117454233A
- Authority
- CN
- China
- Prior art keywords
- data
- layer
- model
- training
- production management
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 91
- 238000007726 management method Methods 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 claims abstract description 54
- 230000008569 process Effects 0.000 claims abstract description 24
- 238000009826 distribution Methods 0.000 claims abstract description 16
- 239000010410 layer Substances 0.000 claims description 93
- 238000012549 training Methods 0.000 claims description 68
- 230000006870 function Effects 0.000 claims description 33
- 238000013461 design Methods 0.000 claims description 22
- 210000002569 neuron Anatomy 0.000 claims description 18
- 238000007781 pre-processing Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 13
- 238000005070 sampling Methods 0.000 claims description 13
- 238000006243 chemical reaction Methods 0.000 claims description 12
- 238000004140 cleaning Methods 0.000 claims description 12
- 230000004927 fusion Effects 0.000 claims description 12
- 238000011176 pooling Methods 0.000 claims description 12
- 230000007613 environmental effect Effects 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 239000006185 dispersion Substances 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 238000005286 illumination Methods 0.000 claims description 3
- 239000011229 interlayer Substances 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- NRNCYVBFPDDJNE-UHFFFAOYSA-N pemoline Chemical compound O1C(N)=NC(=O)C1C1=CC=CC=C1 NRNCYVBFPDDJNE-UHFFFAOYSA-N 0.000 claims description 3
- 238000005096 rolling process Methods 0.000 claims description 3
- 238000011425 standardization method Methods 0.000 claims description 3
- 238000002759 z-score normalization Methods 0.000 claims description 2
- 230000009471 action Effects 0.000 description 3
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Manufacturing & Machinery (AREA)
- Probability & Statistics with Applications (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a safety production management method and system based on positioning identification. The invention belongs to the technical field of intelligent production management, in particular to a safe production management method and system based on positioning identification, wherein the scheme adopts a method for designing a search direction based on definition importance parameters and influence parameters, designing an initial step length, and estimating the stability of the step length to finally obtain a variable step length; through designing inertial weights based on normal distribution, diversity of searching is maintained in the searching process, and step iteration strategies and movement strategies are designed, so that searching is more efficient and accurate, the convergence speed of an algorithm is improved, and a search result is judged based on an adaptability threshold value.
Description
Technical Field
The invention relates to the technical field of intelligent production management, in particular to a safe production management method and system based on positioning identification.
Background
The safety production management method based on positioning identification is a series of measures and methods adopted for guaranteeing the life safety of staff and the safety of working environment and preventing and reducing accidents; however, the general safety production management model has the problems that the model training efficiency and the convergence rate are low due to the poor searching direction, and the training stability is weak due to the too large or too small model step length; the general searching method has the problems that the searching range is limited, so that the searching capability is weak, the algorithm convergence speed is low, and the local optimum is easily trapped.
Disclosure of Invention
Aiming at the problems of low model training efficiency and low convergence speed caused by poor search direction and weak training stability caused by overlarge or undersize model step length of a general safety production management model, the method designs the search direction based on definition importance parameters and influence parameters, designs an initial step length and carries out stability estimation on the step length to finally obtain a variable step length, so that the model better balances updating amplitude and stability and improves model performance; aiming at the problems that the searching range of a general searching method is limited, so that the searching capability is weak, the algorithm convergence speed is low and is easy to fall into local optimum, the scheme maintains the searching diversity in the searching process by designing the inertia weight based on normal distribution, designs the step iteration strategy and the movement strategy, so that the searching is more efficient and accurate, the algorithm convergence speed is improved, and the searching result is judged based on the adaptability threshold.
The technical scheme adopted by the invention is as follows: the invention provides a safety production management method based on positioning identification, which comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: establishing a safe production management model, defining a search direction based on defining importance parameters and influence parameters, and performing stability estimation on the step length based on the design initial step length, so as to obtain a final variable step length, and finishing updating of bias and weight;
step S4: model parameter searching, namely completing the design of a movement strategy based on the design of an inertia weight and step length iteration strategy based on normal distribution, and judging a search result based on an adaptability threshold;
step S5: and (5) safe production management.
Further, in step S1, the data acquisition is to acquire historical data, acquire a positioning address mark, acquire production data, environmental data and a risk level based on a positioning card, and each area is provided with a positioning base station, and the positioning base station acquires area data and collects the area data to a positioning system; the positioning address mark is the position information of the staff and comprises longitude, latitude, altitude, floor number and area number; the production data comprises the number of products, the production rate, the production time and the running state of equipment; the environmental data comprise temperature, humidity, air pressure, illumination intensity and air quality indexes; the risk level includes normal safety, primary production anomaly, secondary production anomaly and tertiary production anomaly.
Further, in step S2, the data preprocessing is to perform data cleaning, vector conversion and normalization processing on the data collected by the positioning system, and the risk level is used as a data tag;
the data cleaning is to perform missing value processing on the collected data, detect missing values in the data and process the missing values by filling the mean value, the median and the mode; performing outlier processing, defining a threshold value of each dimension data, and deleting the data exceeding the threshold value;
the vector conversion is to perform single-heat coding treatment on the data type after data cleaning to serve as a characteristic dimension, and the data size serves as a value of the data dimension;
the normalization process is to normalize the data of each dimension by using Z-ScoreLine normalization processing using the formula z s =(x s - μ)/σ; wherein z is s Is the normalized value, x s Is the value before normalization processing, i.e. after vector conversion, μ is the average value of the data, σ is the standard deviation of the data; the data are converted into a data set with standard normal distribution by using a Z-Score standardization method, seven components of the data set are randomly selected to be used as a training set, three components are used as a test set, a training set is utilized to train a model, and the test set is used for verifying the performance of the model.
Further, in step S3, the secure production management model is established by establishing a neural network, the convolution kernel has a size of 3×1, the pooling kernel has a size of 2×1, and the upsampling size is 2×1; the model consists of six parts: an input layer, a downsampling layer, an upsampling layer, a fusion layer, a jump connection layer and an output layer; the downsampling layer comprises a rolling layer and a pooling layer; firstly, continuously extracting the characteristics of data through a downsampling layer, and gradually restoring the data to the original size through upsampling; pixels are classified one by one, and the fusion layer and the jump connection layer prevent over fitting; the data firstly enters a convolution layer from an input layer, the convolution layer adopts all-zero filling, the number of convolution kernels is set to be 64, and a ReLU activation function is used; features are extracted twice through convolution operation, and the result is sent to a pooling layer, a discarding layer is added in the convolution layer, 50% of neurons are discarded to prevent overfitting; the downsampling layer repeats the process for five times and continuously extracts data characteristics; the convolution operation process is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein X is t+1，I Is the output of the I-th neuron at layer t+1, X t，I Is the output of the ith neuron at layer t, W t，J Is the weight, B t，J Is the offset, pad (·) is the all-zero padding function, is the convolution operation, and f (·) is the activation function;
data enters an up-sampling layer from a discarding layer; the function of the up-sampling layer is opposite to that of the pooling layer, the up-sampling layer doubles the data length, the number of convolution kernels is halved, in order to prevent over-fitting, interlayer residual jump operation is used, data is output through the matrix overlapped by the fusion layer and the up-sampling layer, and model output is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein X is t+1 Is the output of all neurons at layer t+1, X t Is the output of all neurons at layer t, + is the matrix superposition, merge (·) is the fusion function, X t-sn Is the output of all neurons at the t-sn layer;
the loss function of the model adopts a binary cross entropy function, and the calculation formula is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Where Loss is a Loss function, NY is the number of samples, IY is the sample index, Y IY Is a real label->Is a label for model prediction; a set of gradient training parameters is designed, and the gradient parameter correction method specifically comprises the following steps:
step S31: defining importance parameters for measuring the importance of the previous gradient to the current search direction, wherein the following formula is used:
；
wherein beta is t Is the importance parameter of the t-th training, g t Is the gradient of the T-th training, T is the current training time, T is the transpose, y t-1 Is an approximately diagonal term of the Hessian matrix,is the norm of the t-1 th training gradient;
step S32: defining influence parameters for adjusting the influence degree of the gradient of the previous step on the current search direction, wherein the following formula is used:
；
in θ t Is an importance parameter of the t-th training, m t-1 Is the search direction of the t-1 th training;
step S33: the search direction is defined using the following formula:
；
wherein m is t Is the search direction of the t-th training;
step S34: the initial step size is designed by the following formula:
；
wherein alpha is t Is the initial step length of the t-th training, alpha t-1 Is the initial step of the t-1 th training,is a super parameter;
step S35: the stability estimation is performed on the initial step size using the following formula:
；
in the method, in the process of the invention,is an estimate of the stability of the initial step size at the t-th training,is the stability estimation of the initial step length in the t-1 th training;
step S36: the final variable step size is calculated using the following formula:
；
wherein alpha is t ' is the final variable step of the t-th training, alpha is the learning rate, t max The maximum training times are shown, and epsilon is a minimum value for preventing denominator from being 0;
step S37: the weights and offsets are updated using the following formula:
；
；
in which W is t And B t Weights and biases for the t-th training, W t-1 And B t-1 The weights and offsets for the t-1 th training, respectively.
Further, in step S4, the model parameter search specifically includes the following steps:
step S41: initializing, namely randomly initializing a parameter position based on a search space, taking the performance of a neural network model established based on the parameter position as an individual fitness value of the parameter position, wherein a formula for initializing the parameter position is as follows:
；
wherein x is i+1，j Is the position of the (i+1) th body in the j dimension, x i，j Is the position of the ith individual in the j dimension, r i A random number from 0 to 1;
step S42: the inertial weight omega based on normal distribution is designed, and the formula is as follows:
；
wherein omega is min Is the minimum inertial weight, ω max Is the maximum inertial weight, n is the current iteration number, n max Is the maximum number of iterations that can be performed,is a parameter representing the degree of dispersion of the parameter location;
step S43: the step iteration strategy is designed, and the formula is as follows:
；
in delta n Is the step size, delta of the nth iteration max Is the maximum step size, delta min Is the maximum step length, r 1 And r 2 Is a random number of 0 to 1 independent of each other, sigma is a parameter for measuring the deviation of a random variable from a desired value, delta s The search step is accurate;
step S44: the movement strategy is designed using the following formula:
；
in the method, in the process of the invention,is the position of individual i dimension j in the n+1th iteration,/o>Is the position of the i individual in the nth iteration of the j dimension, sign () is a sign function, f () is a fitness value function, +.>Is the position of the i+1 individual in the nth iteration of dimension j,/i>Is the position of the i-1 individual in the nth iteration of dimension j;
step S45: searching and judging, presetting an fitness threshold value, and if the fitness value of an individual is higher than the fitness threshold value, establishing a safe production management model based on the parameter position; if the maximum iteration times are reached, reinitializing the parameter positions for searching; otherwise, continuing the iterative search.
Further, in step S5, the safety production management searches for an initial parameter of the model optimum in step S4, trains the model according to the training set after data acquisition and data preprocessing in step S3, and completes the model establishment if the loss function of the model to the training set is lower than the loss threshold; and acquiring a positioning address mark, production data and environmental data in real time, outputting a danger level by the model, presetting a danger level threshold, and giving early warning to staff if the danger level output by the model is higher than the danger level threshold.
The invention provides a safety production management system based on positioning identification, which comprises a data acquisition module, a data preprocessing module, a safety production management model building module, a model parameter searching module and a safety production management module;
the data acquisition module acquires a positioning address mark, production data, environmental data and danger level based on a positioning card, each area is provided with a positioning base station, the positioning base stations acquire area data and gather the area data to a positioning system, and the data are sent to the data preprocessing module;
the data preprocessing module performs data cleaning, vector conversion and normalization processing on the data acquired by the positioning system, takes the dangerous grade as a data tag, and sends the data to the safety production management model building module;
the safety production management model defines a search direction based on defining importance parameters and influence parameters, performs stability estimation on the step length based on designing an initial step length, thereby obtaining a final variable step length, finishing updating bias and weight, and transmitting data to a model parameter search module;
the model parameter searching module completes the design of a movement strategy based on the design of inertia weight and step length iteration strategy based on normal distribution, judges a searching result based on an adaptability threshold value and sends data to the safety production management module;
the safety production management module carries out safety management on production work based on data collected by the positioning system in real time and a safety production management model.
By adopting the scheme, the beneficial effects obtained by the invention are as follows:
(1) Aiming at the problems that the model training efficiency and the convergence speed are low due to poor search direction and the training stability is weak due to overlarge or undersize model step length in a general safety production management model, the scheme designs the search direction based on definition importance parameters and influence parameters, designs initial step length and carries out stability estimation on the step length, and finally obtains variable step length, so that the model better balances updating amplitude and stability and improves model performance.
(2) Aiming at the problems that the searching range of a general searching method is limited, so that the searching capability is weak, the algorithm convergence speed is low and is easy to fall into local optimum, the scheme maintains the searching diversity in the searching process by designing the inertia weight based on normal distribution, designs the step iteration strategy and the movement strategy, so that the searching is more efficient and accurate, the algorithm convergence speed is improved, and the searching result is judged based on the adaptability threshold.
Drawings
FIG. 1 is a flow chart of a safety production management method based on positioning identification provided by the invention;
FIG. 2 is a schematic diagram of a secure production management system based on location identification provided by the present invention;
FIG. 3 is a flow chart of step S3;
fig. 4 is a flow chart of step S4.
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention; all other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," "outer," and the like indicate orientation or positional relationships based on those shown in the drawings, merely to facilitate description of the invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention.
Referring to fig. 1, the method for safety production management based on positioning identification provided by the invention comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: establishing a safe production management model, defining a search direction based on defining importance parameters and influence parameters, and performing stability estimation on the step length based on the design initial step length, so as to obtain a final variable step length, and finishing updating of bias and weight;
step S4: model parameter searching, namely completing the design of a movement strategy based on the design of an inertia weight and step length iteration strategy based on normal distribution, and judging a search result based on an adaptability threshold;
step S5: and (5) safe production management.
In step S1, the data acquisition is to acquire historical data, and acquire a positioning address mark, production data, environment data and danger level based on a positioning card, wherein each area is provided with a positioning base station, and the positioning base station acquires area data and collects the area data to a positioning system; the positioning address mark is the position information of the staff and comprises longitude, latitude, altitude, floor number and area number; the production data comprises the number of products, the production rate, the production time and the running state of equipment; the environmental data comprise temperature, humidity, air pressure, illumination intensity and air quality indexes; the risk level includes normal safety, primary production anomaly, secondary production anomaly and tertiary production anomaly.
In step S2, the data preprocessing is to perform data cleaning, vector conversion and normalization processing on the data collected by the positioning system, and the risk level is used as a data tag, referring to fig. 1;
the data cleaning is to perform missing value processing on the collected data, detect missing values in the data and process the missing values by filling the mean value, the median and the mode; performing outlier processing, defining a threshold value of each dimension data, and deleting the data exceeding the threshold value;
the vector conversion is to perform single-heat coding treatment on the data type after data cleaning to serve as a characteristic dimension, and the data size serves as a value of the data dimension;
the normalization process is to normalize the data of each dimension by using Z-Score normalization, wherein the formula is Z s =(x s - μ)/σ; wherein z is s Is the normalized value, x s Is the value before normalization processing, i.e. after vector conversion, μ is the average value of the data, σ is the standard deviation of the data; the data are converted into a data set with standard normal distribution by using a Z-Score standardization method, seven components of the data set are randomly selected to be used as a training set, three components are used as a test set, a training set is utilized to train a model, and the test set is used for verifying the performance of the model.
In step S3, a neural network is established, the size of the convolution kernel is 3×1, the size of the pooling kernel is 2×1, and the upsampling size is 2×1; the model consists of six parts: an input layer, a downsampling layer, an upsampling layer, a fusion layer, a jump connection layer and an output layer; the downsampling layer comprises a rolling layer and a pooling layer; firstly, continuously extracting the characteristics of data through a downsampling layer, and gradually restoring the data to the original size through upsampling; pixels are classified one by one, and the fusion layer and the jump connection layer prevent over fitting; the data firstly enters a convolution layer from an input layer, the convolution layer adopts all-zero filling, the number of convolution kernels is set to be 64, and a ReLU activation function is used; features are extracted twice through convolution operation, and the result is sent to a pooling layer, a discarding layer is added in the convolution layer, 50% of neurons are discarded to prevent overfitting; the downsampling layer repeats the process for five times and continuously extracts data characteristics; the convolution operation process is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein X is t+1，I Is the output of the I-th neuron at layer t+1, X t，I Is the output of the ith neuron at layer t, W t，J Is the weight, B t，J Is the offset, pad (·) is the all-zero padding function, is the convolution operation, and f (·) is the activation function;
data enters an up-sampling layer from a discarding layer; the function of the up-sampling layer is opposite to that of the pooling layer, the up-sampling layer doubles the data length, the number of convolution kernels is halved, in order to prevent over-fitting, interlayer residual jump operation is used, data is output through the matrix overlapped by the fusion layer and the up-sampling layer, and model output is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein X is t+1 Is the output of all neurons at layer t+1, X t Is the output of all neurons at layer t, + is the matrix superposition, merge (·) is the fusion function, X t-sn Is the output of all neurons at the t-sn layer;
the loss function of the model adopts a binary cross entropy function, and the calculation formula is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Where Loss is a Loss function, NY is the number of samples, IY is the sample index, Y IY Is a real label->Is a label for model prediction; a set of gradient training parameters is designed, and the gradient parameter correction method specifically comprises the following steps:
step S31: defining importance parameters for measuring the importance of the previous gradient to the current search direction, wherein the following formula is used:
；
wherein beta is t Is the importance parameter of the t-th training, g t Is the gradient of the T-th training, T is the current training time, T is the transpose, y t-1 Is an approximately diagonal term of the Hessian matrix,is the norm of the t-1 th training gradient;
step S32: defining influence parameters for adjusting the influence degree of the gradient of the previous step on the current search direction, wherein the following formula is used:
；
in θ t Is an importance parameter of the t-th training, m t-1 Is the search direction of the t-1 th training;
step S33: the search direction is defined using the following formula:
；
wherein m is t Is the search direction of the t-th training;
step S34: the initial step size is designed by the following formula:
；
wherein alpha is t Is the initial step length of the t-th training, alpha t-1 Is the initial step of the t-1 th training,is a super parameter;
step S35: the stability estimation is performed on the initial step size using the following formula:
；
in the method, in the process of the invention,is an estimate of the stability of the initial step size at the t-th training,is the stability estimation of the initial step length in the t-1 th training;
step S36: the final variable step size is calculated using the following formula:
；
wherein alpha is t ' is the final variable step of the t-th training, alpha is the learning rate, t max The maximum training times are shown, and epsilon is a minimum value for preventing denominator from being 0;
step S37: the weights and offsets are updated using the following formula:
；
；
in which W is t And B t Weights and biases for the t-th training, W t-1 And B t-1 The weights and offsets for the t-1 th training, respectively.
By executing the operation, the method and the device solve the problems that the model training efficiency and the convergence speed are low due to poor search direction and the training stability is weak due to overlarge or undersize model step length in a general safety production management model.
Fifth embodiment referring to fig. 1 and 4, based on the above embodiment, in step S4, the model initial parameter search specifically includes the following steps:
step S41: initializing, namely randomly initializing a parameter position based on a search space, taking the performance of a neural network model established based on the parameter position as an individual fitness value of the parameter position, wherein a formula for initializing the parameter position is as follows:
；
wherein x is i+1，j Is the position of the (i+1) th body in the j dimension, x i，j Is the position of the ith individual in the j dimension, r i A random number from 0 to 1;
step S42: the inertial weight omega based on normal distribution is designed, and the formula is as follows:
；
wherein omega is min Is the minimum inertial weight, ω max Is the maximum inertial weight, n is the current iteration number, n max Is the maximum number of iterations that can be performed,is a parameter representing the degree of dispersion of the parameter location;
step S43: the step iteration strategy is designed, and the formula is as follows:
；
in delta n Is the step size, delta of the nth iteration max Is the maximum step size, delta min Is the maximum step length, r 1 And r 2 Is a random number of 0 to 1 independent of each other, sigma is a parameter for measuring the deviation of a random variable from a desired value, delta s The search step is accurate;
step S44: the movement strategy is designed using the following formula:
；
in the method, in the process of the invention,is the position of individual i dimension j in the n+1th iteration,/o>Is the position of the i individual in the nth iteration of the j dimension, sign () is a sign function, f () is a fitness value function, +.>Is the position of the i+1 individual in the nth iteration of dimension j,/i>Is the position of the i-1 individual in the nth iteration of dimension j;
step S45: searching and judging, presetting an fitness threshold value, and if the fitness value of an individual is higher than the fitness threshold value, establishing a safe production management model based on the parameter position; if the maximum iteration times are reached, reinitializing the parameter positions for searching; otherwise, continuing the iterative search.
By executing the operation, aiming at the problems that the general searching method has weak searching capability due to limited searching range and low algorithm convergence speed and is easy to fall into local optimum, the scheme maintains searching diversity in the searching process by designing inertial weights based on normal distribution, designs a step iteration strategy and a movement strategy so that the searching is more efficient and accurate, improves the algorithm convergence speed and judges the searching result based on an adaptability threshold value.
In step S5, the safety production management searches for the initial parameters of the model optimization by using step S4, trains the model by using step S3 according to the training set after data acquisition and data preprocessing, and completes the model establishment if the loss function of the model to the training set is lower than the loss threshold value, referring to fig. 1; and acquiring a positioning address mark, production data and environmental data in real time, outputting a danger level by the model, presetting a danger level threshold, and giving early warning to staff if the danger level output by the model is higher than the danger level threshold.
An embodiment seven, referring to fig. 2, based on the above embodiment, the safety production management system based on positioning identification provided by the invention includes a data acquisition module, a data preprocessing module, a safety production management model establishment module, a model parameter search module and a safety production management module;
the data acquisition module acquires a positioning address mark, production data, environmental data and danger level based on a positioning card, each area is provided with a positioning base station, the positioning base stations acquire area data and gather the area data to a positioning system, and the data are sent to the data preprocessing module;
the data preprocessing module performs data cleaning, vector conversion and normalization processing on the data acquired by the positioning system, takes the dangerous grade as a data tag, and sends the data to the safety production management model building module;
the safety production management model defines a search direction based on defining importance parameters and influence parameters, performs stability estimation on the step length based on designing an initial step length, thereby obtaining a final variable step length, finishing updating bias and weight, and transmitting data to a model parameter search module;
the model parameter searching module completes the design of a movement strategy based on the design of inertia weight and step length iteration strategy based on normal distribution, judges a searching result based on an adaptability threshold value and sends data to the safety production management module;
the safety production management module carries out safety management on production work based on data collected by the positioning system in real time and a safety production management model.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
The invention and its embodiments have been described above with no limitation, and the actual construction is not limited to the embodiments of the invention as shown in the drawings. In summary, if one of ordinary skill in the art is informed by this disclosure, a structural manner and an embodiment similar to the technical solution should not be creatively devised without departing from the gist of the present invention.
Claims (7)
1. The safe production management method based on positioning identification is characterized by comprising the following steps of: the method comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: establishing a safe production management model, defining a search direction based on defining importance parameters and influence parameters, and performing stability estimation on the step length based on the design initial step length, so as to obtain a final variable step length, and finishing updating of bias and weight;
step S4: model parameter searching, namely completing the design of a movement strategy based on the design of an inertia weight and step length iteration strategy based on normal distribution, and judging a search result based on an adaptability threshold;
step S5: safety production management;
in step S3, the safety production management model is established by establishing a neural network, the size of the convolution kernel is 3×1, the size of the pooling kernel is 2×1, and the up-sampling size is 2×1; the model consists of six parts: an input layer, a downsampling layer, an upsampling layer, a fusion layer, a jump connection layer and an output layer; the downsampling layer comprises a rolling layer and a pooling layer; firstly, continuously extracting the characteristics of data through a downsampling layer, and gradually restoring the data to the original size through upsampling; pixels are classified one by one, and the fusion layer and the jump connection layer prevent over fitting; the data firstly enters a convolution layer from an input layer, the convolution layer adopts all-zero filling, the number of convolution kernels is set to be 64, and a ReLU activation function is used; features are extracted twice through convolution operation, and the result is sent to a pooling layer, a discarding layer is added in the convolution layer, 50% of neurons are discarded to prevent overfitting; the downsampling layer repeats the process for five times and continuously extracts data characteristics; the convolution operation process is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein X is t+1，I Is the output of the I-th neuron at layer t+1, X t，I Is the output of the ith neuron at layer t, W t，J Is the weight, B t，J Is the offset, pad (·) is the all-zero padding function, is the convolution operation, and f (·) is the activation function;
data enters an up-sampling layer from a discarding layer; the function of the up-sampling layer is opposite to that of the pooling layer, the up-sampling layer doubles the data length, the number of convolution kernels is halved, in order to prevent over-fitting, interlayer residual jump operation is used, data is output through the matrix overlapped by the fusion layer and the up-sampling layer, and model output is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein X is t+1 Is the output of all neurons at layer t+1, X t Is the output of all neurons at layer t, + is the matrix superposition, merge (·) is the fusion function, X t-sn Is the output of all neurons at the t-sn layer;
the loss function of the model adopts a binary cross entropy function, and the calculation formula is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Where Loss is a Loss function, NY is the number of samples, IY is the sample index, Y IY Is a real label->Is a label for model prediction; a set of gradient training parameters is designed, and the gradient parameter correction method specifically comprises the following steps:
step S31: defining importance parameters for measuring the importance of the previous gradient to the current search direction, wherein the following formula is used:
；
wherein beta is t Is the importance parameter of the t-th training, g t Is the gradient of the T-th training, T is the current training time, T is the transpose, y t-1 Is an approximately diagonal term of the Hessian matrix,is the norm of the t-1 th training gradient;
step S32: defining influence parameters for adjusting the influence degree of the gradient of the previous step on the current search direction, wherein the following formula is used:
；
in θ t Is an importance parameter of the t-th training, m t-1 Is the search direction of the t-1 th training;
step S33: the search direction is defined using the following formula:
；
wherein m is t Is the search direction of the t-th training;
step S34: the initial step size is designed by the following formula:
；
wherein alpha is t Is the initial step length of the t-th training, alpha t-1 Is the initial step of the t-1 th training,is a super parameter;
step S35: the stability estimation is performed on the initial step size using the following formula:
；
in the method, in the process of the invention,is the stability estimate for the initial step at the time of training t,/, for>Is the stability estimation of the initial step length in the t-1 th training;
step S36: the final variable step size is calculated using the following formula:
；
wherein alpha is t ' is the final variable step of the t-th training, alpha is the learning rate, t max The maximum training times are shown, and epsilon is a minimum value for preventing denominator from being 0;
step S37: the weights and offsets are updated using the following formula:
；
；
in which W is t And B t Weights and biases for the t-th training, W t-1 And B t-1 The weights and offsets for the t-1 th training, respectively.
2. The location identification based secure production management method according to claim 1, wherein: in step S4, the model parameter search specifically includes the following steps:
step S41: initializing, namely randomly initializing a parameter position based on a search space, taking the performance of a neural network model established based on the parameter position as an individual fitness value of the parameter position, wherein a formula for initializing the parameter position is as follows:
；
wherein x is i+1，j Is the position of the (i+1) th body in the j dimension, x i，j Is the position of the ith individual in the j dimension, r i A random number from 0 to 1;
step S42: the inertial weight omega based on normal distribution is designed, and the formula is as follows:
；
wherein omega is min Is the minimum inertial weight, ω max Is the maximum inertial weight, n is the current iteration number, n max Is the maximum number of iterations that can be performed,is a parameter representing the degree of dispersion of the parameter location;
step S43: the step iteration strategy is designed, and the formula is as follows:
；
in delta n Is the step size, delta of the nth iteration max Is the maximum step size, delta min Is the maximum step length, r 1 And r 2 Is a random number of 0 to 1 independent of each other, sigma is a parameter for measuring the deviation of a random variable from a desired value, delta s The search step is accurate;
step S44: the movement strategy is designed using the following formula:
；
in the method, in the process of the invention,is the position of individual i dimension j in the n+1th iteration,/o>Is the position of the i individual in the nth iteration of the j dimension, sign () is a sign function, f () is a fitness value function, +.>Is the position of the i+1 individual in the nth iteration of dimension j,/i>Is the position of the i-1 individual in the nth iteration of dimension j;
step S45: searching and judging, presetting an fitness threshold value, and if the fitness value of an individual is higher than the fitness threshold value, establishing a safe production management model based on the parameter position; if the maximum iteration times are reached, reinitializing the parameter positions for searching; otherwise, continuing the iterative search.
3. The location identification based secure production management method according to claim 1, wherein: in step S1, the data acquisition is to acquire historical data, acquire a positioning address mark, acquire production data, environmental data and danger level based on a positioning card, and each area is provided with a positioning base station, and the positioning base station acquires area data and gathers the area data to a positioning system; the positioning address mark is the position information of the staff and comprises longitude, latitude, altitude, floor number and area number; the production data comprises the number of products, the production rate, the production time and the running state of equipment; the environmental data comprise temperature, humidity, air pressure, illumination intensity and air quality indexes; the risk level includes normal safety, primary production anomaly, secondary production anomaly and tertiary production anomaly.
4. The location identification based secure production management method according to claim 1, wherein: in step S2, the data preprocessing is to perform data cleaning, vector conversion and normalization processing on the data collected by the positioning system, and the risk level is used as a data tag;
the data cleaning is to perform missing value processing on the collected data, detect missing values in the data and process the missing values by filling the mean value, the median and the mode; performing outlier processing, defining a threshold value of each dimension data, and deleting the data exceeding the threshold value;
the vector conversion is to perform single-heat coding treatment on the data type after data cleaning to serve as a characteristic dimension, and the data size serves as a value of the data dimension;
the normalization process is to normalize the data of each dimension by using Z-Score normalization, wherein the formula is Z s =(x s - μ)/σ; wherein z is s Is the normalized value, x s Is the value before normalization processing, i.e. after vector conversion, μ is the average value of the data, σ is the standard deviation of the data; the data are converted into a data set with standard normal distribution by using a Z-Score standardization method, seven components of the data set are randomly selected to be used as a training set, three components are used as a test set, a training set is utilized to train a model, and the test set is used for verifying the performance of the model.
5. The location identification based secure production management method according to claim 1, wherein: in step S5, the safety production management searches for the initial parameters of the optimal model in step S4, trains the model according to the training set after data acquisition and data preprocessing in step S3, and completes the model establishment if the loss function of the model to the training set is lower than the loss threshold; and acquiring a positioning address mark, production data and environmental data in real time, outputting a danger level by the model, presetting a danger level threshold, and giving early warning to staff if the danger level output by the model is higher than the danger level threshold.
6. A secure production management system based on location identification for implementing the secure production management method based on location identification according to any one of claims 1 to 5, characterized in that: the system comprises a data acquisition module, a data preprocessing module, a safe production management model building module, a model parameter searching module and a safe production management module.
7. The location identification based secure production management system of claim 6, wherein:
the data acquisition module acquires a positioning address mark, production data, environmental data and danger level based on a positioning card, each area is provided with a positioning base station, the positioning base stations acquire area data and gather the area data to a positioning system, and the data are sent to the data preprocessing module;
the data preprocessing module performs data cleaning, vector conversion and normalization processing on the data acquired by the positioning system, takes the dangerous grade as a data tag, and sends the data to the safety production management model building module;
the safety production management model defines a search direction based on defining importance parameters and influence parameters, performs stability estimation on the step length based on designing an initial step length, thereby obtaining a final variable step length, finishing updating bias and weight, and transmitting data to a model parameter search module;
the model parameter searching module completes the design of a movement strategy based on the design of inertia weight and step length iteration strategy based on normal distribution, judges a searching result based on an adaptability threshold value and sends data to the safety production management module;
the safety production management module carries out safety management on production work based on data collected by the positioning system in real time and a safety production management model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311782147.6A CN117454233B (en) | 2023-12-22 | 2023-12-22 | Safety production management method and system based on positioning identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311782147.6A CN117454233B (en) | 2023-12-22 | 2023-12-22 | Safety production management method and system based on positioning identification |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117454233A true CN117454233A (en) | 2024-01-26 |
| CN117454233B CN117454233B (en) | 2024-03-22 |
Family
ID=89589549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311782147.6A Active CN117454233B (en) | 2023-12-22 | 2023-12-22 | Safety production management method and system based on positioning identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117454233B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118075032A (en) * | 2024-04-19 | 2024-05-24 | 湖南天冠电子信息技术有限公司 | Virtual local area network security risk prediction method based on artificial intelligence |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200371520A1 (en) * | 2019-05-24 | 2020-11-26 | Hefei University Of Technology | Path planning method and system for self-driving of autonomous system |
| CN112926117A (en) * | 2021-03-01 | 2021-06-08 | 郑州大学 | BAS-BP-based pile foundation horizontal bearing capacity prediction method |
| CN115081595A (en) * | 2022-07-12 | 2022-09-20 | 盐城工学院 | Neural network optimization method based on integration of improved longicorn algorithm and bat algorithm |
| CN115456136A (en) * | 2022-08-31 | 2022-12-09 | 北京理工大学 | High-rotation body angular rate online estimation method based on improved sparrow search algorithm under gyroscope-free condition |
| CN116644970A (en) * | 2023-05-10 | 2023-08-25 | 福建工程学院 | Photovoltaic power prediction method based on VMD decomposition and lamination deep learning |
| US20230316069A1 (en) * | 2021-05-24 | 2023-10-05 | Soochow University | Wind speed prediction method based on forward complex-valued neural network |
| CN117037831A (en) * | 2023-08-30 | 2023-11-10 | 太原科技大学 | Voice enhancement method based on multi-domain fusion network low-cost automatic search construction |
| WO2023216383A1 (en) * | 2022-05-13 | 2023-11-16 | 苏州大学 | Complex-valued timing signal prediction method based on complex-valued neural network |
-
2023
- 2023-12-22 CN CN202311782147.6A patent/CN117454233B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200371520A1 (en) * | 2019-05-24 | 2020-11-26 | Hefei University Of Technology | Path planning method and system for self-driving of autonomous system |
| CN112926117A (en) * | 2021-03-01 | 2021-06-08 | 郑州大学 | BAS-BP-based pile foundation horizontal bearing capacity prediction method |
| US20230316069A1 (en) * | 2021-05-24 | 2023-10-05 | Soochow University | Wind speed prediction method based on forward complex-valued neural network |
| WO2023216383A1 (en) * | 2022-05-13 | 2023-11-16 | 苏州大学 | Complex-valued timing signal prediction method based on complex-valued neural network |
| CN115081595A (en) * | 2022-07-12 | 2022-09-20 | 盐城工学院 | Neural network optimization method based on integration of improved longicorn algorithm and bat algorithm |
| CN115456136A (en) * | 2022-08-31 | 2022-12-09 | 北京理工大学 | High-rotation body angular rate online estimation method based on improved sparrow search algorithm under gyroscope-free condition |
| CN116644970A (en) * | 2023-05-10 | 2023-08-25 | 福建工程学院 | Photovoltaic power prediction method based on VMD decomposition and lamination deep learning |
| CN117037831A (en) * | 2023-08-30 | 2023-11-10 | 太原科技大学 | Voice enhancement method based on multi-domain fusion network low-cost automatic search construction |
Non-Patent Citations (1)
| Title |
|---|
| 柳长源: ""基于深度图像的球形果实识别定位算法"", 《农业机械学报》, vol. 53, no. 10, 31 October 2022 (2022-10-31), pages 228 - 232 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118075032A (en) * | 2024-04-19 | 2024-05-24 | 湖南天冠电子信息技术有限公司 | Virtual local area network security risk prediction method based on artificial intelligence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117454233B (en) | 2024-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109522793A (en) | More people's unusual checkings and recognition methods based on machine vision | |
| CN117454233B (en) | Safety production management method and system based on positioning identification | |
| CN107423730B (en) | Human gait behavior active detection and recognition system and method based on semantic folding | |
| CN106384093B (en) | A kind of human motion recognition method based on noise reduction autocoder and particle filter | |
| CN104182772B (en) | A kind of gesture identification method based on deep learning | |
| CN111611874B (en) | Face mask wearing detection method based on ResNet and Canny | |
| CN101563710B (en) | Method and apparatus for identifying properties of an object detected by a video surveillance camera | |
| CN110135319A (en) | A kind of anomaly detection method and its system | |
| CN105160400A (en) | L21 norm based method for improving convolutional neural network generalization capability | |
| CN108734171A (en) | A kind of SAR remote sensing image ocean floating raft recognition methods of depth collaboration sparse coding network | |
| CN105657402A (en) | Depth map recovery method | |
| CN110532850B (en) | Fall detection method based on video joint points and hybrid classifier | |
| CN107103308A (en) | A kind of pedestrian's recognition methods again learnt based on depth dimension from coarse to fine | |
| CN106991666A (en) | A kind of disease geo-radar image recognition methods suitable for many size pictorial informations | |
| CN114463675B (en) | Underwater fish group activity intensity identification method and device | |
| CN108762503A (en) | A kind of man-machine interactive system based on multi-modal data acquisition | |
| CN102880870B (en) | The extracting method of face characteristic and system | |
| CN117436707B (en) | Fire safety management method and system based on artificial intelligence | |
| CN107967941A (en) | A kind of unmanned plane health monitoring method and system based on intelligent vision reconstruct | |
| CN109492700A (en) | A kind of Target under Complicated Background recognition methods based on multidimensional information fusion | |
| CN112560948A (en) | Eye fundus map classification method and imaging method under data deviation | |
| CN110163157A (en) | A method of more people's Attitude estimations are carried out using novel loss function | |
| CN114626419B (en) | Action recognition method based on channel state information in WIFI and improved convolutional neural network | |
| CN116152678A (en) | Marine disaster-bearing body identification method based on twin neural network under small sample condition | |
| CN115995040A (en) | SAR image small sample target recognition method based on multi-scale network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |