CN117520753B - Early warning system and method for ice and snow sports - Google Patents
Early warning system and method for ice and snow sports Download PDFInfo
- Publication number
- CN117520753B CN117520753B CN202410017954.7A CN202410017954A CN117520753B CN 117520753 B CN117520753 B CN 117520753B CN 202410017954 A CN202410017954 A CN 202410017954A CN 117520753 B CN117520753 B CN 117520753B
- Authority
- CN
- China
- Prior art keywords
- hidden layer
- early warning
- individual
- output
- data
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 99
- 210000002569 neuron Anatomy 0.000 claims abstract description 144
- 230000035945 sensitivity Effects 0.000 claims abstract description 77
- 238000005457 optimization Methods 0.000 claims abstract description 22
- 230000000739 chaotic effect Effects 0.000 claims abstract description 16
- 238000007781 pre-processing Methods 0.000 claims abstract description 16
- 238000013507 mapping Methods 0.000 claims abstract description 10
- 238000012217 deletion Methods 0.000 claims abstract description 9
- 230000037430 deletion Effects 0.000 claims abstract description 9
- 238000003780 insertion Methods 0.000 claims abstract description 9
- 230000037431 insertion Effects 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 84
- 230000006870 function Effects 0.000 claims description 43
- 230000008569 process Effects 0.000 claims description 41
- 210000004027 cell Anatomy 0.000 claims description 9
- 238000004140 cleaning Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- MYVIATVLJGTBFV-UHFFFAOYSA-M thiamine(1+) chloride Chemical compound [Cl-].CC1=C(CCO)SC=[N+]1CC1=CN=C(C)N=C1N MYVIATVLJGTBFV-UHFFFAOYSA-M 0.000 claims description 3
- 238000004260 weight control Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims 2
- 230000002159 abnormal effect Effects 0.000 claims 1
- 230000006978 adaptation Effects 0.000 claims 1
- 230000009471 action Effects 0.000 description 3
- 238000010845 search algorithm Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000002349 favourable effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 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
- G06F18/10—Pre-processing; Data cleansing
-
- 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
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- 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
- G06N3/048—Activation functions
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an early warning system and method for ice and snow sports, the system includes: the system comprises a data acquisition module, a data preprocessing module, a danger early warning model building module, a danger early warning model parameter searching module and a real-time danger early warning module. The invention belongs to the technical field of dangerous early warning, in particular to an early warning system and method for ice and snow sports, wherein a dangerous early warning model module is constructed to introduce indirect sensitivity and direct sensitivity to obtain the total sensitivity of hidden layer neurons, the insertion or deletion of the hidden layer neurons is determined based on a loss function and the total sensitivity, and the number of hidden layer neurons of a dangerous early warning model is determined; dangerous early warning model parameter search module uses Circle chaotic mapping to initialize individual position based on random number delta i And determining an updating strategy of the individual, generating a opposition position to perform position selection, performing position optimization, and determining model parameters according to the global optimal position.
Description
Technical Field
The invention belongs to the technical field of danger early warning, and particularly relates to an early warning system and method for ice and snow sports.
Background
The early warning system for ice and snow sports is used for processing relevant data of ice and snow sports so as to prevent dangerous accidents in advance and protect the safety of athletes. However, the optimal number of hidden layer neurons in the existing early warning system is difficult to determine, and the hidden layer neurons are unevenly distributed, so that the problems of low model training efficiency and over-fitting are caused; the existing parameter search algorithm has the problems that the initialization position is uneven, the convergence speed is low, local optimum is easily trapped, global optimum solution cannot be found, and the performance of a model is reduced.
Disclosure of Invention
In view of the above, the present invention provides a pre-apparatus for ice and snow sportsAiming at the problems that the optimal number of hidden layer neurons is difficult to determine, the weight distribution of the hidden layer neurons is uneven, and the model training efficiency is low and the model is over-fitted in the existing early warning system, the scheme introduces indirect sensitivity and direct sensitivity to obtain the total sensitivity of the hidden layer neurons, determines the insertion or deletion of the hidden layer neurons based on a loss function and the total sensitivity, determines the number of the hidden layer neurons of a dangerous early warning model, completes the training of the model, dynamically adjusts the weight of the hidden layer neurons, thereby improving the performance and effect of the model and enhancing the training efficiency and convergence; aiming at the problems that the existing parameter search algorithm has uneven initialization positions and low convergence speed, is easy to fall into local optimum, cannot find global optimum solutions and causes the performance of a model to be reduced, the scheme uses Circle chaotic mapping to initialize individual positions, is favorable for the diversity of the individual positions and is based on random number delta i And determining an updating strategy of an individual, generating opposite positions to perform position selection, expanding the searching range, improving the exploration capability of an algorithm, simultaneously avoiding sinking into local optimization, performing position optimization, determining model parameters according to the global optimal position, and improving the precision of parameter searching.
The technical scheme adopted by the invention is as follows: the invention provides an early warning system for ice and snow sports, which comprises a data acquisition module, a data preprocessing module, a danger early warning model building module, a danger early warning model parameter searching module and a real-time danger early warning module, wherein the data acquisition module is used for acquiring data of ice and snow sports;
the data acquisition module acquires relevant data of ice and snow sports and corresponding classification labels, and sends the data to the data preprocessing module;
the data preprocessing module performs vector conversion, data cleaning and normalization processing on the collected data and sends the data to the module for constructing the dangerous early warning model;
the method comprises the steps that an indirect sensitivity and a direct sensitivity are introduced by a danger early warning model building module to obtain the total sensitivity of hidden layer neurons, insertion or deletion of the hidden layer neurons is determined based on a loss function and the total sensitivity, the number of hidden layer neurons of a danger early warning model is determined, training of the model is completed, and data are sent to a danger early warning model parameter searching module;
the dangerous early warning model parameter searching module uses Circle chaotic mapping to initialize individual positions and is based on random numbersDetermining an updating strategy of an individual, generating a opposition position to perform position selection, performing position optimization, determining model parameters according to the global optimal position, and transmitting data to a real-time danger early warning module;
the real-time danger early warning module inputs the data acquired in real time into the danger early warning model for classification, and real-time early warning is carried out based on the classification label output by the danger early warning model.
The invention provides an early warning method for ice and snow sports, which comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: constructing a dangerous early warning model, introducing indirect sensitivity and direct sensitivity to obtain total sensitivity of hidden layer neurons, determining insertion or deletion of the hidden layer neurons based on a loss function and the total sensitivity, determining the number of the hidden layer neurons of the dangerous early warning model, and completing training of the model;
step S4: dangerous early warning model parameter search, initializing individual positions by using Circle chaotic mapping, and based on random numbersDetermining an updating strategy of an individual, generating a opposition position to perform position selection, performing position optimization, and determining model parameters according to the global optimal position;
step S5: real-time hazard pre-warning.
Further, in step S1, the data acquisition is to acquire relevant data of the ice and snow sports and a corresponding classification label, wherein the corresponding classification label includes normal and dangerous types, and the acquired data is used as sample data.
Further, in step S2, the data preprocessing is to perform vector conversion, data cleaning and normalization processing on the collected data, and the data cleaning is to process missing values, outliers and noise in the data.
Further, in step S3, the construction of the hazard early warning model specifically includes the following steps:
step S31: the method comprises the steps of designing a framework, wherein a danger early warning model comprises an input layer, an output layer and a hidden layer, the input layer receives input data, the output layer carries out final classified output, the hidden layer carries out feature extraction and information transmission, the hidden layer consists of neurons, and the neurons comprise a memory unit, an input gate, a forgetting gate and an output gate;
step S32: initializing a network, randomly initializing the number of hidden layer neurons and network parameters, presetting the maximum training times G and a loss threshold zeta (n), wherein n is a training time index;
step S33: the loss function is calculated using the formula:
；
where E (N) is a loss function, N K Is the number of sample data that is to be processed,classification tag, which is the model output,>is a true classification label, d is a sample data index;
step S34: the indirect sensitivity of hidden layer neurons is calculated as follows:
step S341: the indirect output of the neuron is calculated using the following formula:
；
in the method, in the process of the invention,is the indirect output of the h hidden layer neuron in the nth training, and is +.>Is the indirect output of the h hidden layer neuron in the n-1 th training, B x 、B l 、B f And B o Weighted control signals of input gate, memory cell, forget gate and output gate, q x 、q l 、q f And q o Bias of input gate, memory cell, forget gate and output gate, respectively, h is hidden layer neuron index, s (·), g (·), f (·) and z (·) are activation functions of input gate, memory cell, forget gate and output gate, respectively, x n Is the input data of the nth training;
step S342: the indirect output of the output layer is calculated using the following formula:
；
in the method, in the process of the invention,is the indirect output of the output layer in the nth training, N H Is the number of neurons in the hidden layer,is the weight of the h hidden layer neuron connected to the output layer during the nth training;
step S343: the indirect sensitivity was calculated using the formula:
；
in the method, in the process of the invention,is the indirect sensitivity of the h hidden layer neuron in the nth training, var [ & gt]Is a variance function->Is->Corresponding loss function, ++>Is an indirect conditional constraint, < ->Is->Is a function of the variance of (a),is at->Under the condition->Is a variance of (2);
step S35: the direct sensitivity of hidden layer neurons is calculated as follows:
step S351: the direct output of the neuron is calculated using the following formula:
；
in the method, in the process of the invention,is the direct output of the h hidden layer neuron in the nth training, and is +.>Is the direct output of the h hidden layer neuron in the n-1 th training;
step S352: the direct output of the output layer is calculated using the following formula:
；
in the method, in the process of the invention,is the direct output of the output layer during the nth training;
step S353: the direct sensitivity was calculated using the following formula:
；
in the method, in the process of the invention,is the direct sensitivity of the h hidden layer neuron in the nth training, and +.>Is->Corresponding loss function, ++>Is a direct conditional constraint, ->Is->Is a function of the variance of (a),is at->Under the condition->Is a variance of (2);
step S36: the total sensitivity was calculated using the following formula:
；
in the method, in the process of the invention,is the total sensitivity of the h hidden layer neuron in the nth training;
step S37: model training, the steps are as follows:
step S371: inserting a new hidden layer neuron, if E (n) > ζ (n), inserting the new hidden layer neuron, initializing the weight of the new hidden layer neuron, and the formula for initializing the weight of the new hidden layer neuron is as follows:
；
；
；
；
；
；
in the method, in the process of the invention,、/>、/>and->Respectively, the n-th training is input into the gate, the memory unit and the memory unitForgetting gate and output gate weight control signal for new hidden layer neurons, +.>Is the weight of the new hidden layer neuron output from the feedback loop at the nth training, +.>Is the weight of the new hidden layer neuron connected to the output layer in the nth training, w is the hidden layer neuron index with the highest total sensitivity, < ->、/>、And->The input gate, the memory unit, the forgetting gate and the output gate are respectively used for weighting control signals of hidden layer neurons with highest total sensitivity in the nth training, and the hidden layer neurons are respectively used for receiving the signals of the hidden layer neurons with highest total sensitivity in the nth training>Is the weight of the hidden layer neuron with the highest total sensitivity output from the feedback loop in the nth training, +.>Is the weight of the hidden layer neuron with the highest total sensitivity connected to the output layer during the nth training;
step S372: deleting neurons of the hidden layer if<ζ (n), deleting the h hidden layer neuron, setting the weight of the deleted hidden layer neuron to 0, and updating the connection weight between the remaining hidden layer neurons and the output layer;
step S38: the model is determined, a loss function E (n) is updated, and when the loss function E (n) is less than or equal to a loss threshold value zeta (n), the number of hidden layer neurons of the dangerous early warning model is determined; otherwise, if the maximum training times G are reached, the network is reinitialized; otherwise, model training is performed again.
Further, in step S4, the risk early warning model parameter search specifically includes the following steps:
step S41: initializing the position of the individual, representing the position of the individual by using model parameters, and randomly generating an initial value Q 0 The individual locations are initialized using Circle chaotic map, using the following formula:
；
in which Q i Is the location of the i-th individual,is the position of the i-1 th individual, i is the individual index, N Q Is the total number of individuals, sin (·) is a sine function, mod (·) is a remainder function, ub and lb are the upper and lower limits of the individual search space, respectively; a and b are parameters of Circle chaotic map for controlling the shape of the map;
step S42: calculating a fitness value, and taking model performance established based on model parameters as an individual fitness value;
step S43: location update, generating a random number for each individualBased on random number->The update strategy of the individual is determined using the following formula:
；
；
in the method, in the process of the invention,is the position of the ith individual after the position update, Z is the exploration factor, cos (·) is the cosine function, r 1 、r 2 、r 3 And r 4 Is 4 mutually independent random numbers, u is a first constant, T is an iteration number index, T is a maximum iteration number, p is a first random number, W is a second random number, Q A And Q S Two individual positions selected randomly;
step S44: the position is changed, and the steps are as follows:
step S441: the opposite position is generated using the following formula:
；
in the method, in the process of the invention,is->M is a scale factor;
step S442: the location is selected using the following formula:
；
in the method, in the process of the invention,is the position after the i-th individual selection at the t-th iteration, < >>Is->Is adapted to the value of->Is->Is a fitness value of (a);
step S443: calculating a global optimal position, updating the fitness value of the individual, and selecting the individual position with the highest fitness value as the global optimal position Q best ；
Step S444: position optimization, generating a random number for each dimension of each individualBased on random number->The optimization strategy of the individual is determined by the following formula:
；
；
in the method, in the process of the invention,is the position of the ith individual after the jth dimension optimization at the t-th iteration, j is the dimension index, N D Is the dimension of the individual search space,/->Is the preliminary optimized position of the ith individual in the jth dimension at the t-th iteration, c is a third random number, e is a second constant, beta max And beta min Probability maximum and probability minimum, respectively;
step S45: parameter determination, namely presetting an fitness value evaluation threshold alpha, updating the fitness value and the global optimal position of an individual, and constructing a danger early warning model based on model parameters when the fitness value corresponding to the global optimal position is higher than the fitness value evaluation threshold alpha; otherwise, if the maximum iteration number T is reached, the individual position is reinitialized; otherwise, the location update is performed again.
Further, in step S5, the real-time risk early warning is to collect data of the athlete during the ice and snow sports in real time, input the data into the risk early warning model for classification after the data is preprocessed, and perform real-time early warning on the risk of the athlete during the ice and snow sports based on the classification label output by the risk early warning model.
By adopting the scheme, the beneficial effects obtained by the invention are as follows:
(1) Aiming at the problems that the optimal number of hidden layer neurons is difficult to determine, the weight distribution of the hidden layer neurons is uneven, and the model training efficiency is low and the model is over-fitted in the existing early warning system, the scheme introduces indirect sensitivity and direct sensitivity to obtain the total sensitivity of the hidden layer neurons, the insertion or deletion of the hidden layer neurons is determined based on a loss function and the total sensitivity, the number of the hidden layer neurons of the dangerous early warning model is determined, the training of the model is completed, the weight of the hidden layer neurons is dynamically adjusted, and therefore the performance and the effect of the model are improved, and the training efficiency and the convergence are enhanced.
(2) Aiming at the problems that the existing parameter search algorithm has uneven initialization positions and low convergence speed, is easy to fall into local optimum, cannot find global optimum solutions and causes the performance of a model to be reduced, the scheme uses Circle chaotic mapping to initialize individual positions, is favorable for the diversity of the individual positions and is based on random number delta i And determining an updating strategy of an individual, generating opposite positions to perform position selection, expanding the searching range, improving the exploration capability of an algorithm, simultaneously avoiding sinking into local optimization, performing position optimization, determining model parameters according to the global optimal position, and improving the precision of parameter searching.
Drawings
FIG. 1 is a schematic diagram of an early warning system for ice and snow sports provided by the invention;
FIG. 2 is a schematic flow chart of an early warning method for ice and snow sports provided by the 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 early warning system for ice and snow sports provided by the invention comprises a data acquisition module, a data preprocessing module, a danger early warning model building module, a danger early warning model parameter searching module and a real-time danger early warning module;
the data acquisition module acquires relevant data of ice and snow sports and corresponding classification labels, and sends the data to the data preprocessing module;
the data preprocessing module performs vector conversion, data cleaning and normalization processing on the collected data and sends the data to the module for constructing the dangerous early warning model;
the method comprises the steps that an indirect sensitivity and a direct sensitivity are introduced by a danger early warning model building module to obtain the total sensitivity of hidden layer neurons, insertion or deletion of the hidden layer neurons is determined based on a loss function and the total sensitivity, the number of hidden layer neurons of a danger early warning model is determined, training of the model is completed, and data are sent to a danger early warning model parameter searching module;
the dangerous early warning model parameter searching module uses Circle chaotic mapping to initialize individual positions and is based on random numbersDetermining an updating strategy of an individual, generating a opposition position to perform position selection, performing position optimization, determining model parameters according to the global optimal position, and transmitting data to a real-time danger early warning module;
the real-time danger early warning module inputs the data acquired in real time into the danger early warning model for classification, and real-time early warning is carried out based on the classification label output by the danger early warning model.
Referring to fig. 2, the early warning method for ice and snow sports provided by the present invention includes the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: constructing a dangerous early warning model, introducing indirect sensitivity and direct sensitivity to obtain total sensitivity of hidden layer neurons, determining insertion or deletion of the hidden layer neurons based on a loss function and the total sensitivity, determining the number of the hidden layer neurons of the dangerous early warning model, and completing training of the model;
step S4: dangerous early warning model parameter search, initializing individual positions by using Circle chaotic mapping, and based on random numbersDetermining an updating strategy of an individual, generating a opposition position to perform position selection, performing position optimization, and determining model parameters according to the global optimal position;
step S5: real-time hazard pre-warning.
In step S1, data collection is to collect relevant data of ice and snow sports and corresponding classification labels, where the corresponding classification labels include normal and dangerous types, and the collected data is used as sample data, referring to fig. 2.
Fourth embodiment referring to fig. 2, the embodiment is based on the above embodiment, and in step S2, the data preprocessing is to perform vector conversion, data cleaning and normalization processing on the collected data, and the data cleaning is to process missing values, outliers and noise in the data.
In a fifth embodiment, referring to fig. 2 and 3, the embodiment is based on the above embodiment, and in step S3, the construction of the hazard early warning model specifically includes the following steps:
step S31: the method comprises the steps of designing a framework, wherein a danger early warning model comprises an input layer, an output layer and a hidden layer, the input layer receives input data, the output layer carries out final classified output, the hidden layer carries out feature extraction and information transmission, the hidden layer consists of neurons, and the neurons comprise a memory unit, an input gate, a forgetting gate and an output gate;
step S32: initializing a network, randomly initializing the number of hidden layer neurons and network parameters, and presetting the maximum training times G and a loss threshold value ζ (n) =n -0.65 N is the training frequency index;
step S33: the loss function is calculated using the formula:
；
where E (N) is a loss function, N K Is the number of sample data that is to be processed,classification tag, which is the model output,>is a true classification label, d is a sample data index;
step S34: the indirect sensitivity of hidden layer neurons is calculated as follows:
step S341: the indirect output of the neuron is calculated using the following formula:
；
in the method, in the process of the invention,is the indirect output of the h hidden layer neuron in the nth training, and is +.>Is the indirect output of the h hidden layer neuron in the n-1 th training, B x 、B l 、B f And B o Weighted control signals of input gate, memory cell, forget gate and output gate, q x 、q l 、q f And q o Bias of input gate, memory cell, forget gate and output gate, respectively, h is hidden layer neuron index, s (·), g (·), f (·) and z (·) are activation functions of input gate, memory cell, forget gate and output gate, respectively, x n Is the input data of the nth training;
step S342: the indirect output of the output layer is calculated using the following formula:
；
in the method, in the process of the invention,is the indirect output of the output layer in the nth training, N H Is the number of neurons in the hidden layer,is the weight of the h hidden layer neuron connected to the output layer during the nth training;
step S343: the indirect sensitivity was calculated using the formula:
；
in the method, in the process of the invention,is the indirect sensitivity of the h hidden layer neuron in the nth training, var [ & gt]Is a variance function->Is->Corresponding loss function, ++>Is an indirect conditional constraint, < ->Is->Is a function of the variance of (a),is at->Under the condition->Is a variance of (2);
step S35: the direct sensitivity of hidden layer neurons is calculated as follows:
step S351: the direct output of the neuron is calculated using the following formula:
；
in the method, in the process of the invention,is the direct output of the h hidden layer neuron in the nth training, and is +.>Is the direct output of the h hidden layer neuron in the n-1 th training;
step S352: the direct output of the output layer is calculated using the following formula:
；
in the method, in the process of the invention,is the direct output of the output layer during the nth training;
step S353: the direct sensitivity was calculated using the following formula:
；
in the method, in the process of the invention,is the direct sensitivity of the h hidden layer neuron in the nth training, and +.>Is->Corresponding loss function, ++>Is a direct conditional constraint, ->Is->Is a function of the variance of (a),is at->Under the condition->Is a variance of (2);
step S36: the total sensitivity was calculated using the following formula:
；
in the method, in the process of the invention,is the total sensitivity of the h hidden layer neuron in the nth training;
step S37: model training, the steps are as follows:
step S371: inserting a new hidden layer neuron, if E (n) > ζ (n), inserting the new hidden layer neuron, initializing the weight of the new hidden layer neuron, and the formula for initializing the weight of the new hidden layer neuron is as follows:
；
；
；
；
；
；
in the method, in the process of the invention,、/>、/>and->The input gate, the memory unit, the forgetting gate and the output gate respectively carry out weight control signals of the new hidden layer neuron during the nth training, and the +.>Is the weight of the new hidden layer neuron output from the feedback loop at the nth training, +.>Is the weight of the new hidden layer neuron connected to the output layer in the nth training, w is the hidden layer neuron index with the highest total sensitivity, < ->、/>、And->The input gate, the memory unit, the forgetting gate and the output gate are respectively used for weighting control signals of hidden layer neurons with highest total sensitivity in the nth training, and the hidden layer neurons are respectively used for receiving the signals of the hidden layer neurons with highest total sensitivity in the nth training>Is the weight of the hidden layer neuron with the highest total sensitivity output from the feedback loop in the nth training, +.>Is the weight of the hidden layer neuron with the highest total sensitivity connected to the output layer during the nth training;
step S372: deleting neurons of the hidden layer if<ζ (n), deleting the h hidden layer neuron, setting the weight of the deleted hidden layer neuron to 0, and updating the connection weight between the remaining hidden layer neurons and the output layer;
step S38: the model is determined, a loss function E (n) is updated, and when the loss function E (n) is less than or equal to a loss threshold value zeta (n), the number of hidden layer neurons of the dangerous early warning model is determined; otherwise, if the maximum training times G are reached, the network is reinitialized; otherwise, model training is performed again.
By executing the operation, the problems that the optimal number of hidden layer neurons is difficult to determine, the weight distribution of the hidden layer neurons is uneven, and the model training efficiency is low and the model is over-fitted are solved for the existing early warning system.
In a sixth embodiment, referring to fig. 2 and 4, based on the above embodiment, in step S4, the risk early warning model parameter search specifically includes the following steps:
step S41: initializing the individual position, representing the individual position by model parameters, at [0,1]Within-range random generation of initial value Q 0 The individual locations are initialized using Circle chaotic map, using the following formula:
；
in which Q i Is the location of the i-th individual,is the position of the i-1 th individual, i is the individual index, N Q Is the total number of individuals, sin (·) is a sine function, mod (·) is a remainder functionUb and lb are the upper and lower limits of the individual search space, respectively; a and b are parameters of Circle chaotic map for controlling the shape of the map;
step S42: calculating a fitness value, and taking model performance established based on model parameters as an individual fitness value;
step S43: location update, generating one [0,1 ] for each individual]Random numbers within a rangeBased on random number->The update strategy of the individual is determined using the following formula:
；
；
in the method, in the process of the invention,is the position of the ith individual after the position update, Z is the exploration factor, cos (·) is the cosine function, r 1 、r 2 、r 3 And r 4 Is->4 mutually independent random numbers in the range, u is +.>The first constant in the range, T is the iteration number index, T is the maximum iteration number, and p is [ -1,1]A first random number in the range, W is [ -Z, Z]Second random number in range, Q A And Q S Two individual positions selected randomly;
step S44: the position is changed, and the steps are as follows:
step S441: the opposite position is generated using the following formula:
；
in the method, in the process of the invention,is->M is a scale factor;
step S442: the location is selected using the following formula:
；
in the method, in the process of the invention,is the position after the i-th individual selection at the t-th iteration, < >>Is->Is adapted to the value of->Is->Is a fitness value of (a);
step S443: calculating a global optimal position, updating the fitness value of the individual, and selecting the individual position with the highest fitness value as the global optimal position Q best ；
Step S444: position optimization, generating one for each dimension of each individualRandom number within range->Based on random number->The optimization strategy of the individual is determined by the following formula:
；
；
in the method, in the process of the invention,is the position of the ith individual after the jth dimension optimization at the t-th iteration, j is the dimension index, N D Is the dimension of the individual search space,/->Is the preliminary optimized position of the ith individual in the jth dimension at the t-th iteration, c isA third random number in the range, e is +.>Second constant within the range, beta max And beta min Probability maximum and probability minimum, respectively;
step S45: parameter determination, namely presetting an fitness value evaluation threshold alpha, updating the fitness value and the global optimal position of an individual, and constructing a danger early warning model based on model parameters when the fitness value corresponding to the global optimal position is higher than the fitness value evaluation threshold alpha; otherwise, if the maximum iteration number T is reached, the individual position is reinitialized; otherwise, the location update is performed again.
By executing the operation, aiming at the problems that the existing parameter searching algorithm has uneven initialization position, slow convergence speed, easy sinking into local optimum and inability to find global optimum solution, resulting in reduced model performance, the scheme uses Circle chaotic mapping to initialize individual positions,is beneficial to the diversity of the individual positions and is based on the random number delta i And determining an updating strategy of an individual, generating opposite positions to perform position selection, expanding the searching range, improving the exploration capability of an algorithm, simultaneously avoiding sinking into local optimization, performing position optimization, determining model parameters according to the global optimal position, and improving the precision of parameter searching.
In step S5, the real-time hazard pre-warning is to collect the data of the athlete during the ice and snow sports in real time, input the data into the hazard pre-warning model for classification after the data is pre-processed, and pre-warn the hazards of the athlete during the ice and snow sports in real time based on the classification label output by the hazard pre-warning model, as shown in fig. 2.
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 (6)
1. A early warning system for ice and snow sports, its characterized in that: the system comprises a data acquisition module, a data preprocessing module, a danger early warning model building module, a danger early warning model parameter searching module and a real-time danger early warning module;
the data acquisition module acquires relevant data of ice and snow sports and corresponding classification labels, and sends the data to the data preprocessing module;
the data preprocessing module performs vector conversion, data cleaning and normalization processing on the collected data and sends the data to the module for constructing the dangerous early warning model;
the method comprises the steps that an indirect sensitivity and a direct sensitivity are introduced by a danger early warning model building module to obtain the total sensitivity of hidden layer neurons, insertion or deletion of the hidden layer neurons is determined based on a loss function and the total sensitivity, the number of hidden layer neurons of a danger early warning model is determined, training of the model is completed, and data are sent to a danger early warning model parameter searching module;
the dangerous early warning model parameter searching module uses Circle chaotic mapping to initialize individual positions and is based on random number delta i Determining an updating strategy of an individual, generating a opposition position to perform position selection, performing position optimization, determining model parameters according to the global optimal position, and transmitting data to a real-time danger early warning module;
the real-time danger early warning module inputs the data acquired in real time into the danger early warning model for classification, and carries out real-time early warning based on the classification label output by the danger early warning model;
in step S3, the construction of the hazard early warning model specifically includes the following steps:
step S31: the method comprises the steps of designing a framework, wherein a danger early warning model comprises an input layer, an output layer and a hidden layer, the input layer receives input data, the output layer carries out final classified output, the hidden layer carries out feature extraction and information transmission, the hidden layer consists of neurons, and the neurons comprise a memory unit, an input gate, a forgetting gate and an output gate;
step S32: initializing a network, randomly initializing the number of hidden layer neurons and network parameters, presetting the maximum training times G and a loss threshold zeta (n), wherein n is a training time index;
step S33: the loss function is calculated using the formula:
；
where E (N) is a loss function, N K Is the number of sample data that is to be processed,classification tag, which is the model output,>is a true classification label, d is a sample data index;
step S34: calculating the indirect sensitivity of hidden layer neurons;
step S35: calculating the direct sensitivity of hidden layer neurons;
step S36: the total sensitivity was calculated using the following formula:
；
in the method, in the process of the invention,is the total sensitivity of the h hidden layer neuron in the nth training;
step S37: training a model;
step S38: the model is determined, a loss function E (n) is updated, and when the loss function E (n) is less than or equal to a loss threshold value zeta (n), the number of hidden layer neurons of the dangerous early warning model is determined; otherwise, if the maximum training times G are reached, the network is reinitialized; otherwise, model training is conducted again;
in step S34, the calculating the indirect sensitivity of the hidden layer neuron specifically includes the steps of:
step S341: the indirect output of the neuron is calculated using the following formula:
；
in the method, in the process of the invention,is the indirect output of the h hidden layer neuron in the nth training, and is +.>Is the indirect output of the h hidden layer neuron in the n-1 th training, B x 、B l 、B f And B o Weighted control signals of input gate, memory cell, forget gate and output gate, q x 、q l 、q f And q o Bias of input gate, memory cell, forget gate and output gate, respectively, h is hidden layer neuron index, s (·), g (·), f (·) and z (·) are activation functions of input gate, memory cell, forget gate and output gate, respectively, x n Is the input data of the nth training;
step S342: the indirect output of the output layer is calculated using the following formula:
；
in the method, in the process of the invention,is the indirect output of the output layer in the nth training, N H Is the number of neurons in the hidden layer, +.>Is the weight of the h hidden layer neuron connected to the output layer during the nth training;
step S343: the indirect sensitivity was calculated using the formula:
；
in the method, in the process of the invention,is the indirect sensitivity of the h hidden layer neuron in the nth training, var [ & gt]Is the function of the variance of the signal,is->Corresponding loss function, ++>Is an indirect conditional constraint, < ->Is->Is a function of the variance of (a),is at->Under the condition->Is a variance of (2);
in step S35, the calculating the direct sensitivity of the hidden layer neuron specifically includes the steps of:
step S351: the direct output of the neuron is calculated using the following formula:
；
in the method, in the process of the invention,is the direct output of the h hidden layer neuron in the nth training, and is +.>Is the direct output of the h hidden layer neuron in the n-1 th training;
step S352: the direct output of the output layer is calculated using the following formula:
；
in the method, in the process of the invention,is the direct output of the output layer during the nth training;
step S353: the direct sensitivity was calculated using the following formula:
；
in the method, in the process of the invention,is the direct sensitivity of the h hidden layer neuron in the nth training, and +.>Is->Corresponding loss function, ++>Is a direct conditional constraint, ->Is->Variance of->Is atUnder the condition->Is a variance of (c).
2. An early warning method for ice and snow sports, which is applied to the early warning system for ice and snow sports described in the above claim 1, is characterized in that: the method comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: constructing a dangerous early warning model, introducing indirect sensitivity and direct sensitivity to obtain total sensitivity of hidden layer neurons, determining insertion or deletion of the hidden layer neurons based on a loss function and the total sensitivity, determining the number of the hidden layer neurons of the dangerous early warning model, and completing training of the model;
step S4: dangerous early warning model parameter search, initializing individual positions by using Circle chaotic mapping, and based on random number delta i Determining an updating strategy of an individual, generating a opposition position to perform position selection, performing position optimization, and determining model parameters according to the global optimal position;
step S5: real-time hazard pre-warning.
3. The early warning method for ice and snow sports according to claim 2, characterized in that: in step S37, the model training specifically includes the steps of:
step S371: inserting a new hidden layer neuron, if E (n) > ζ (n), inserting the new hidden layer neuron, initializing the weight of the new hidden layer neuron, and the formula for initializing the weight of the new hidden layer neuron is as follows:
；
；
；
；
；
；
in the method, in the process of the invention,、/>、/>and->The input gate, the memory unit, the forgetting gate and the output gate respectively carry out weight control signals of the new hidden layer neuron during the nth training, and the +.>Is the weight of the new hidden layer neuron output from the feedback loop at the nth training, +.>Is the weight of the new hidden layer neuron connected to the output layer in the nth training, w is the hidden layer neuron index with the highest total sensitivity, < ->、/>、/>Andthe input gate, the memory unit, the forgetting gate and the output gate are respectively used for weighting control signals of hidden layer neurons with highest total sensitivity in the nth training, and the hidden layer neurons are respectively used for receiving the signals of the hidden layer neurons with highest total sensitivity in the nth training>Is the weight of the hidden layer neuron with the highest total sensitivity output from the feedback loop in the nth training, +.>Is the weight of the hidden layer neuron with the highest total sensitivity connected to the output layer during the nth training;
step S372: deleting neurons of the hidden layer if<ζ (n), deleting the h hidden layer neuron, setting the weight of the deleted hidden layer neuron to 0, and updating the connection weights between the remaining hidden layer neurons and the output layer.
4. The early warning method for ice and snow sports according to claim 2, characterized in that: in step S4, the dangerous early warning model parameter searching specifically includes the following steps:
step S41: initializing the position of the individual, representing the position of the individual by using model parameters, and randomly generating an initial value Q 0 The individual locations are initialized using Circle chaotic map, using the following formula:
；
in which Q i Is the location of the i-th individual,is the position of the i-1 th individual, i is the individual index, N Q Is the total number of individuals, sin (·) is a sine function, mod (·) is a remainder function, ub and lb are the upper and lower limits of the individual search space, respectively; a and b are parameters of Circle chaotic map;
step S42: calculating a fitness value, and taking model performance established based on model parameters as an individual fitness value;
step S43: location update, generating a random number delta for each individual i Based on random number delta i The update strategy of the individual is determined using the following formula:
；
；
in the method, in the process of the invention,is the position of the ith individual after the position update, Z is the exploration factor, cos (·) is the cosine function, r 1 、r 2 、r 3 And r 4 Is 4 mutually independent random numbers, u is a first constant, T is an iteration number index, and T is the maximum iteration numberThe number, p, is a first random number, W is a second random number, Q A And Q S Two individual positions selected randomly;
step S44: position transformation;
step S45: parameter determination, namely presetting an fitness value evaluation threshold alpha, updating the fitness value and the global optimal position of an individual, and constructing a danger early warning model based on model parameters when the fitness value corresponding to the global optimal position is higher than the fitness value evaluation threshold alpha; otherwise, if the maximum iteration number T is reached, the individual position is reinitialized; otherwise, the location update is performed again.
5. The method for pre-warning ice and snow sports according to claim 4, wherein: in step S44, the position transformation specifically includes the steps of:
step S441: the opposite position is generated using the following formula:
；
in the method, in the process of the invention,is Q i The opposite position of (t), m being a scale factor;
step S442: the location is selected using the following formula:
；
in the method, in the process of the invention,is the position after the i-th individual selection at the t-th iteration, < >>Is->Is used for the adaptation value of the (a),is->Is a fitness value of (a);
step S443: calculating a global optimal position, updating the fitness value of the individual, and selecting the individual position with the highest fitness value as the global optimal position Q best ；
Step S444: position optimization, generating a random number ε for each dimension of each individual i，j Based on random number epsilon i，j The optimization strategy of the individual is determined by the following formula:
；
；
in the method, in the process of the invention,is the position of the ith individual after the jth dimension optimization at the t-th iteration, j is the dimension index, N D Is the dimension of the individual search space,/->Is the preliminary optimized position of the ith individual in the jth dimension at the t-th iteration, c is a third random number, e is a second constant, beta max And beta min The probability maximum and probability minimum, respectively.
6. The early warning method for ice and snow sports according to claim 2, characterized in that: in step S1, the data acquisition is to acquire relevant data of ice and snow sports and corresponding classification labels, wherein the corresponding classification labels comprise normal and dangerous types, and the acquired data is used as sample data;
in step S2, the data preprocessing is to perform vector conversion, data cleaning and normalization processing on the collected data, and the data cleaning is to process missing values, abnormal values and noise in the data;
in step S5, the real-time hazard pre-warning is to collect the data of the athlete during the ice and snow sports in real time, input the data into the hazard pre-warning model for classification after the data is preprocessed, and pre-warn the hazards of the athlete during the ice and snow sports in real time based on the classification label output by the hazard pre-warning model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410017954.7A CN117520753B (en) | 2024-01-05 | 2024-01-05 | Early warning system and method for ice and snow sports |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410017954.7A CN117520753B (en) | 2024-01-05 | 2024-01-05 | Early warning system and method for ice and snow sports |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117520753A CN117520753A (en) | 2024-02-06 |
| CN117520753B true CN117520753B (en) | 2024-04-05 |
Family
ID=89764914
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410017954.7A Active CN117520753B (en) | 2024-01-05 | 2024-01-05 | Early warning system and method for ice and snow sports |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117520753B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013104938A2 (en) * | 2012-01-11 | 2013-07-18 | 77 Elektronika Muszeripari Kft. | Neural network and a method for teaching thereof |
| CN115438583A (en) * | 2022-09-13 | 2022-12-06 | 合肥城市云数据中心股份有限公司 | Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm |
| CN116990883A (en) * | 2023-09-27 | 2023-11-03 | 北京中科太赫兹科技有限公司 | Remote person-carried dangerous object detection system based on multi-spectrum sensing fusion technology |
| CN117313795A (en) * | 2023-10-09 | 2023-12-29 | 山东电工电气集团数字科技有限公司 | Intelligent building energy consumption prediction method based on improved DBO-LSTM |
-
2024
- 2024-01-05 CN CN202410017954.7A patent/CN117520753B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013104938A2 (en) * | 2012-01-11 | 2013-07-18 | 77 Elektronika Muszeripari Kft. | Neural network and a method for teaching thereof |
| CN115438583A (en) * | 2022-09-13 | 2022-12-06 | 合肥城市云数据中心股份有限公司 | Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm |
| CN116990883A (en) * | 2023-09-27 | 2023-11-03 | 北京中科太赫兹科技有限公司 | Remote person-carried dangerous object detection system based on multi-spectrum sensing fusion technology |
| CN117313795A (en) * | 2023-10-09 | 2023-12-29 | 山东电工电气集团数字科技有限公司 | Intelligent building energy consumption prediction method based on improved DBO-LSTM |
Non-Patent Citations (1)
| Title |
|---|
| 面向工程全局优化的混沌优化算法研究进展;刘振军;杨迪雄;;计算力学学报;20160615(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117520753A (en) | 2024-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111353542A (en) | Training method and device of image classification model, computer equipment and storage medium | |
| CN114386324A (en) | Ultra-short-term wind power segmented prediction method based on turning period identification | |
| CN107679368A (en) | PET/CT high dimensional feature level systems of selection based on genetic algorithm and varied precision rough set | |
| CN108711103A (en) | Personal loan repays Risk Forecast Method, device, computer equipment and medium | |
| CN109582813A (en) | A kind of search method, device, equipment and the storage medium of historical relic showpiece | |
| KR20210095671A (en) | Image processing method and related device | |
| CN117077671B (en) | Interactive data generation method and system | |
| CN117520753B (en) | Early warning system and method for ice and snow sports | |
| CN111144462B (en) | Unknown individual identification method and device for radar signals | |
| CN116469561A (en) | Breast cancer survival prediction method based on deep learning | |
| CN108090507A (en) | A kind of medical imaging textural characteristics processing method based on integrated approach | |
| Meng et al. | PaDE-NPC: Parameter adaptive differential evolution with novel parameter control for single-objective optimization | |
| CN117077041B (en) | Intelligent community management method and system based on Internet of things | |
| CN101739680A (en) | Method for partitioning medical volume data based on artificial endocrine-immune system | |
| Matthews et al. | An improved storm cell identification and tracking (SCIT) algorithm based on DBSCAN clustering and JPDA tracking methods | |
| CN116644205A (en) | Geographic information data processing system and method | |
| CN104346520B (en) | A kind of Data Dimensionality Reduction system and its dimension reduction method based on neuroid | |
| CN113159194A (en) | Missing value filling method based on attribute dynamic selection and gray level correlation analysis | |
| Mohan et al. | Integration of IRS-1A L2 data by fuzzy logic approaches for landuse classification | |
| CN117669008B (en) | Foundation settlement prediction method and system based on deep learning | |
| CN110334811A (en) | Dynamic dataflow classification method based on Bayesian network | |
| CN114511822B (en) | Daily sundries sales system capable of predicting sales volume according to monitoring picture | |
| CN109063934A (en) | Combinatorial Optimization result acquisition methods, equipment and readable medium based on artificial intelligence | |
| CN117350409B (en) | Man-machine dialogue model training system based on machine learning | |
| CN118035799B (en) | Filling mining equipment fault diagnosis method and device based on multidimensional data fusion |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240528 Address after: Room A895, Unit 303, No. 130 Shawan Road, Shawan Street, Panyu District, Guangzhou City, Guangdong Province, 510000 Patentee after: Guangdong Ice and Snow Sports Development Co.,Ltd. Country or region after: China Address before: Zone 222/A, Floor 2, Building B, Enterprise Incubation Building, No. 8, Shenkong Road, Zhangjiakou Hi tech Industrial Development Zone, Hebei Province, 075000 Patentee before: Hebei Zhongtishanjian Sports Industry Co.,Ltd. Country or region before: China |