CN111476102A - Safety protection method, central control equipment and computer storage medium - Google Patents
Safety protection method, central control equipment and computer storage medium Download PDFInfo
- Publication number
- CN111476102A CN111476102A CN202010168130.1A CN202010168130A CN111476102A CN 111476102 A CN111476102 A CN 111476102A CN 202010168130 A CN202010168130 A CN 202010168130A CN 111476102 A CN111476102 A CN 111476102A
- Authority
- CN
- China
- Prior art keywords
- security
- acoustic
- layer
- training
- signals
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000003860 storage Methods 0.000 title claims description 13
- 238000012549 training Methods 0.000 claims abstract description 115
- 238000013145 classification model Methods 0.000 claims abstract description 76
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 37
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 238000000605 extraction Methods 0.000 claims description 69
- 238000011176 pooling Methods 0.000 claims description 50
- 238000009432 framing Methods 0.000 claims description 31
- 230000006870 function Effects 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 19
- 238000012795 verification Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 15
- 238000001514 detection method Methods 0.000 claims description 12
- 230000004927 fusion Effects 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 7
- 238000002372 labelling Methods 0.000 claims description 7
- 238000005065 mining Methods 0.000 abstract description 2
- 230000002265 prevention Effects 0.000 abstract 1
- 238000004590 computer program Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 239000013307 optical fiber Substances 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000002790 cross-validation Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 238000012800 visualization Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000009194 climbing Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/047—Probabilistic or stochastic networks
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Signal Processing (AREA)
- Alarm Systems (AREA)
Abstract
The embodiment of the specification discloses a safety protection method, which is applied to central control equipment, wherein the central control equipment is connected with acoustic sensors arranged in the central control equipment and different security areas, and the method comprises the following steps: determining an acoustic signal training sample set, wherein the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events; through the designed one-dimensional convolutional neural network structure, the distributed acoustic vibration sensors in the security area are used for collecting sample sets of various security events and are used for learning and training the algorithm, the algorithm is used for carrying out automatic feature mining on the sensor vibration signals, the security events can be accurately and quickly identified, and safety prevention and control are effectively carried out. And carrying out classified identification on the safety events through the target safety event classification model, and giving an alarm when the safety events with the target types are identified.
Description
Technical Field
The invention relates to the technical field of perimeter safety protection, in particular to a safety protection method, central control equipment and a computer storage medium.
Background
At present, the optical fiber distributed acoustic sensing system is widely applied to a plurality of fields, such as airport security, national security, city security monitoring, monitoring of oil and gas pipelines and railway lines and other peripheral security. The optical fiber distributed acoustic sensing system can acquire vibration signals through an optical cable sensor, and can monitor the vibration signals in real time for a long time by transmitting the vibration types sent by objects in direct or indirect contact. Through data collection, then through algorithm analysis handles and mode identification intelligence study, can discern the interfering signal of different grade type, for example climb the enclosure, pedestrian's walking, interfering signal such as vehicle process, the early warning that carries on that can realize the system implementation plays the effect to preventing invading area early warning monitoring.
The existing common identification methods are various, characteristics are extracted in time domain, frequency domain and other aspects by manual work at present, and identification is carried out by a machine learning method, but the method has the defects that the characteristics are extracted by manual work and are complex, the implementation is difficult, and the accuracy cannot meet the requirements in practical application scenes. In addition, wavelet analysis is adopted to calculate energy ratio as a signal identification characteristic. Some schemes also convert the one-dimensional vibration signals into energy spectrum images to be analyzed by a deep learning technology, and the method has the technical problems of low processing speed, small detection range and the like.
Disclosure of Invention
In view of the above, the present invention has been developed to provide a security method, apparatus, and computer storage medium that overcome or at least partially address the above-identified problems.
In a first aspect, an embodiment of the present invention provides a safety protection method, which is applied to a central control device of a distributed acoustic sensing security system, where the central control device is connected to acoustic sensors arranged in different security areas, and the method includes:
determining an acoustic signal training sample set, wherein the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired through a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1;
performing multi-round training on signals in the training sample set through a one-dimensional convolutional neural network which is set up in advance, wherein the one-dimensional convolutional neural network comprises a primary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer and an M-class Softmax layer which are arranged in sequence, in each round of training, inputting the signals in the training sample set into the primary feature extraction layer for primary feature extraction, compressing the primary extracted features through the maximum pooling layer, inputting the compressed features into the deep feature extraction layer for deep feature extraction, performing average pooling on the compressed features through the global average pooling layer, inputting the full connection layer for fusion, finally classifying through the Softmax layer to obtain a classification result, and calculating the loss value of the classification result and an actual classification label through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training;
and carrying out classified identification on the safety events of the signals acquired by the acoustic sensors in different security areas through the target safety event classification model, and controlling an alarm device to alarm according to a corresponding alarm strategy under the condition that the safety event with the target type is identified.
Optionally, the determining a training sample set of the acoustic signal includes:
acquiring acoustic signals corresponding to various event security events of a first preset duration acquired by acoustic sensors corresponding to different time periods;
and according to a sliding window with second preset time duration, performing framing processing on each obtained acoustic signal in a mode of sliding for third preset time duration each time, labeling corresponding classification labels on the obtained framing signals, and then extracting all framing signals according to a first preset proportion to form the acoustic signal training sample set.
Optionally, the preliminary feature extraction layer includes a first convolution sublayer and a second convolution sublayer that connect gradually, the deep feature extraction layer includes a third convolution sublayer and a fourth convolution sublayer that connect gradually, the full tie layer and the Softmax layer of one-dimensional convolution neural network still are provided with the Dropout layer, through the Dropout layer abandons the processing to the characteristic.
Optionally, verifying the model for updating the network parameters by verifying the signals in the sample set until a convergence condition is reached includes:
and performing verification test by adopting a 10-fold cross verification mode, starting and stopping early if the classification precision on the verification sample set is smaller than the preset precision to obtain the safety event classification model of the training of the current round, and otherwise, outputting the safety event classification model of the training of the current round until complete iteration.
Optionally, the determining a target security event classification model from a plurality of security event classification models obtained through a plurality of rounds of training includes:
after the obtained framing signals are labeled with corresponding classification labels, extracting all the framing signals according to a second preset proportion to form an acoustic signal test sample set;
performing performance evaluation on each safety event classification model based on the acoustic signal test sample set;
and taking the security event classification model with the optimal performance as a target security event classification model.
In a second aspect, an embodiment of the present invention provides a central control device, which is applied to a distributed acoustic sensing security system, where the central control device is connected to acoustic sensors arranged in different security areas, and the central control device includes:
the device comprises a determining unit, a judging unit and a processing unit, wherein the determining unit is used for determining an acoustic signal training sample set, the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired through a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1;
the model training unit is used for performing multi-round training on the signals in the training sample set through a pre-built one-dimensional convolutional neural network, wherein the one-dimensional convolutional neural network comprises a primary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer and an M-classified Softmax layer which are sequentially arranged, in each round of training, the signals in the training sample set are input into the primary feature extraction layer for primary feature extraction, the primary extracted features are compressed through the maximum pooling layer and then input into the deep feature extraction layer for deep feature extraction, the compressed features are averaged and pooled through the global average pooling layer and then input into the full connection layer for fusion, finally the Softmax layer is used for classification to obtain a classification result, and the loss value of the classification result and an actual classification label is calculated through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training;
and the security unit is used for carrying out security event classification and identification on signals acquired by the acoustic sensors in different security areas through the target security event classification model, and controlling the alarm device to alarm according to a corresponding alarm strategy under the condition of identifying the existence of a security event of a target type.
Optionally, the determining unit is specifically configured to:
acquiring acoustic signals corresponding to various event security events of a first preset duration acquired by acoustic sensors corresponding to different time periods;
and according to a sliding window with second preset time duration, performing framing processing on each obtained acoustic signal in a mode of sliding for third preset time duration each time, labeling corresponding classification labels on the obtained framing signals, and then extracting all framing signals according to a first preset proportion to form the acoustic signal training sample set.
In a third aspect, an embodiment of the present invention provides a central control device, which is applied to a distributed acoustic sensing security system, where the central control device includes a processor and a memory:
the memory is for storing a program for performing the method of any one of the preceding first aspects;
the processor is configured to execute programs stored in the memory.
In a fourth aspect, an embodiment of the present invention provides a computer storage medium for storing computer software instructions for a security protection method according to any one of the above first aspects, which includes a program designed to execute the above aspect.
In a fifth aspect, an embodiment of the present invention provides a distributed acoustic sensing security system, including a central control device and acoustic sensors connected to the central control device and disposed in different security areas, where each acoustic sensor of the central control device acquires an acoustic signal corresponding to a corresponding security area and transmits the acoustic signal to the central control device, and the central control device is configured to perform security protection by using the security protection method according to any one of the first aspect.
The above-mentioned at least one technical solution in the embodiment of the present application has at least one or more of the following technical effects:
the embodiment of the invention provides a safety protection method, which is applied to central control equipment, wherein the central control equipment is connected with acoustic sensors arranged in different security areas, the central control equipment determines an acoustic signal training sample set, the sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired by a corresponding acoustic sensor in a distributed acoustic sensing detection system, then, a one-dimensional convolutional neural network built in advance is used for carrying out multi-round training on the signals in the training sample set, in each round of training, the signals in the training sample set are input into a primary feature extraction layer for primary feature extraction, the primary extracted features are compressed by a maximum pooling layer, then input into a deep feature extraction layer for deep feature extraction, and then, after average pooling is carried out on the compressed features by a global average pooling layer, inputting full connection layers for fusion, finally classifying through a Softmax layer to obtain a classification result, calculating the classification result and a loss value of an actual classification label through a loss function, updating network parameters of a one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model of updating the network parameters through signals in a verification sample set until a convergence condition is reached to obtain a security event classification model of each training round, determining a target security event classification model from a plurality of security event classification models obtained through a plurality of training rounds, finally, performing security event classification recognition on signals collected by acoustic sensors in different security areas through the target security event classification model, and controlling an alarm device to alarm according to a corresponding alarm strategy under the condition that the security event with the target type is recognized. The complexity of manually extracting the features can be saved, a good effect can be achieved in a low sampling rate, the processing speed is high, and the recognition rate is superior to that of a manual extraction algorithm.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 is a flow chart of a security method in a first embodiment of the present invention;
FIG. 2 is a detailed flowchart of the model training portion of the security protection method according to the first embodiment of the present invention;
FIG. 3 is a diagram of a one-dimensional convolutional neural network structure in a first embodiment of the present invention;
FIG. 4 is a model training visualization in a first embodiment of the present invention;
FIG. 5 is a diagram of a confusion matrix in a first embodiment of the present invention;
FIG. 6 is a schematic diagram of a central control apparatus according to a second embodiment of the present invention;
fig. 7 is a schematic diagram of a central control device in a third embodiment of the present invention.
Detailed Description
The embodiment discloses a safety protection method, which is applied to central control equipment of a distributed acoustic sensing security system, wherein the central control equipment is connected with acoustic sensors arranged in different security areas, and the method comprises the following steps: determining an acoustic signal training sample set, wherein the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired through a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1; performing multi-round training on signals in the training sample set through a one-dimensional convolutional neural network which is set up in advance, wherein the one-dimensional convolutional neural network comprises a primary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer and an M-class Softmax layer which are arranged in sequence, in each round of training, inputting the signals in the training sample set into the primary feature extraction layer for primary feature extraction, compressing the primary extracted features through the maximum pooling layer, inputting the compressed features into the deep feature extraction layer for deep feature extraction, performing average pooling on the compressed features through the global average pooling layer, inputting the full connection layer for fusion, finally classifying through the Softmax layer to obtain a classification result, and calculating the loss value of the classification result and an actual classification label through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training; and carrying out classified identification on the safety events of the signals acquired by the acoustic sensors in different security areas through the target safety event classification model, and controlling an alarm device to alarm according to a corresponding alarm strategy under the condition that the safety event with the target type is identified.
The technical solutions of the present invention are described in detail below with reference to the drawings and specific embodiments, and it should be understood that the embodiments and specific features of the embodiments of the present application are described in detail in the technical solutions of the present application, but are not limited to the technical solutions of the present application, and the technical features of the embodiments and the specific embodiments of the present application may be combined with each other without conflict.
The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and/or B, may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" herein generally indicates that the former and latter related objects are in an "or" relationship.
Examples
A first embodiment of the present invention provides a safety protection method, which is applied to a central control device of a distributed acoustic sensing security system, where the central control device is connected to acoustic sensors arranged in different security areas, and please refer to fig. 1, where the method includes the following steps:
s101: determining an acoustic signal training sample set, wherein the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired through a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1;
s102: performing multi-round training on signals in the training sample set through a one-dimensional convolutional neural network which is set up in advance, wherein the one-dimensional convolutional neural network comprises a primary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer and an M-class Softmax layer which are arranged in sequence, in each round of training, inputting the signals in the training sample set into the primary feature extraction layer for primary feature extraction, compressing the primary extracted features through the maximum pooling layer, inputting the compressed features into the deep feature extraction layer for deep feature extraction, performing average pooling on the compressed features through the global average pooling layer, inputting the full connection layer for fusion, finally classifying through the Softmax layer to obtain a classification result, and calculating the loss value of the classification result and an actual classification label through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training;
s103: and carrying out classified identification on the safety events of the signals acquired by the acoustic sensors in different security areas through the target safety event classification model, and controlling an alarm device to alarm according to a corresponding alarm strategy under the condition that the safety event with the target type is identified.
Specifically, the embodiment provides a safety protection method for efficiently and accurately identifying a safety event, which is used for solving the problems of low identification accuracy, low identification speed and small detection range of a distributed acoustic sensing system in the current security field. The method in this embodiment is sequentially divided into three parts, namely data preparation, model training, event recognition and alarm execution, and is shown in fig. 2 in detail, wherein only 3 types of event recognition are shown in fig. 2 as an example, and in a specific implementation process, setting may be performed according to actual needs, and the embodiment is not limited herein.
First, step S101 is executed, and data preparation can be specifically realized by the following steps:
acquiring acoustic signals corresponding to various event security events of a first preset duration acquired by acoustic sensors corresponding to different time periods;
and according to a sliding window with second preset time duration, performing framing processing on each obtained acoustic signal in a mode of sliding for third preset time duration each time, labeling corresponding classification labels on the obtained framing signals, and then extracting all framing signals according to a first preset proportion to form the acoustic signal training sample set.
Specifically, in this embodiment, a security area to be detected is deployed first, where the security area may be an enclosure area, an import/export area, and the like, and in a specific implementation, the security area may be set according to an actual need, and here, this embodiment is not limited. For example, guard event acquisition may be performed by distributed acoustic sensors deployed along a railway. Each security protection area is provided with corresponding acoustic sensor in a distributed mode, and the acoustic sensors indirectly or directly sense vibration signals through media such as the ground or a fence. In this embodiment, the acoustic sensor may be an optical fiber sensor, the optical fiber sampling frequency of the optical fiber sensor is 200Hz, the sensor detection takes 20m as a region range as a sensing point, and the data phase is (-pi, pi). The acoustic sensor acquires sensor sensing signals in real time through the lower computer acquisition equipment, the data of all the sensing nodes acquired in the current second is returned once every second, and the acquired data signals are in a waveform sequence. The central control equipment can be connected to a plurality of lower computer acquisition equipment and can perform rapid identification processing, so that a long-distance range can be detected.
And event data acquisition is carried out near the optical fiber sensor, and the acquired signals comprise signals of M types of security events. The signal acquisition time of the M-type security events is divided into morning, afternoon and evening, a plurality of sensor nodes in the distributed acoustic sensing security system are randomly selected in each time period to acquire signals of various security events, the acquisition time is about two minutes each time, the acquisition is carried out alternately, and the acquisition is carried out for multiple times. In this embodiment, the security event may be the aforementioned wall climbing event, pedestrian walking event, vehicle passing event, and the like, and in a specific implementation process, the security event may be set according to an actual need, which is not limited in this embodiment.
After the central control device receives acoustic signals of various events collected by the distributed acoustic sensor, the data is also framed, according to the foregoing example, for 2-minute acoustic signals collected from the acoustic sensor, 5s is used as an event sliding window, 1s is slid each time, a segment of 2-minute data is decomposed into a plurality of data frames, and each data frame is labeled with a classification label of a belonging security event. The ratio of the obtained data frame to the acoustic signal training sample set and the acoustic signal testing sample set is 4: 1. Further, in the subsequent training process, cross validation needs to be performed, and the acoustic signal training sample set is divided into a training set and a cross validation set, wherein the ratio is 4: 1.
After the step S101 is executed, step S102 is executed, and first, a one-dimensional convolutional neural network is constructed, which includes a preliminary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full-link layer, and an M-class Softmax layer, which are sequentially arranged in this embodiment. The primary feature extraction layer comprises a first convolution sublayer and a second convolution sublayer which are sequentially connected, the deep feature extraction layer comprises a third convolution sublayer and a fourth convolution sublayer which are sequentially connected, a Dropout layer is further arranged between the full connection layer of the one-dimensional convolution neural network and the Softmax layer, and the Dropout layer discards the features.
For example, as shown in fig. 3, the initial two convolutional layers with one dimension of 100 × 10 are initial feature extraction layers for extracting initial features such as shallow time domain and frequency domain features, and then a maximum pooling layer with step size of 3 is used to reduce the neural network parameters. Two 160 × 10 convolutional layers are next to be the deep layer feature extraction layers for extracting the features of the deeper layers. Followed by a global max pooling for processing of the final sorted mattes. Since the model is eventually followed by a Full Connection (FC) layer resulting in excessive parameters, a Dropout layer is added to prevent overfitting. And finally, outputting the three classified Softmax layers for outputting the classification results.
Wherein, the calculation formula of Softmax is as follows:
(0<yi<1,∑iyi1), in the formula, yiRepresenting class i output probability, n representing number of classes, ziAnd zjA value representing the output of the neural network.
After the one-dimensional convolutional neural network is constructed, the acoustic signal training sample set needs to be trained for multiple rounds to obtain a security event classification model obtained by each round of training, wherein the security event classification model is an M-class security event classification model.
At the beginning of network training, initializing a parameter theta of the one-dimensional convolutional neural network in a normal distribution mode, wherein the parameter theta is subjected to normal distribution. In fig. 3, the convolution will be described by taking the first convolution sublayer C1 as an example, where C1 is a one-dimensional convolution layer, the number of C1 convolution kernels is R, the convolution length is R, and each convolution kernel needs to be subjected to K-time convolution calculations. Input data is Xi＝{x1,x2,x3,…,xsW is the weight matrix after the initialization parameter of the jth convolution kernelj＝[Wj1,Wj2,Wj3,…，Wjr]Let the bias unit be bjUsing the Re L U function, the convolution calculation formula is as follows:
Convjk＝ReLU([xi,xi+1,…,xi+r]·[Wj1,Wj2,Wj3,…，Wjr]+bj)
in the above equation, i is the ith data point, j is the jth convolution kernel, k is the kth convolution, s is the input data length, r is the magnitude of the convolution kernel.
ReLU＝max(x,0)
Taking the pooling layer P1 as an example, the pooling is selected to be the maximum pooling, and as shown in fig. 3, each kernel in P1 corresponds to a convolution kernel of the convolutional layer C2, the number of convolution kernels is R, the size is P, and each kernel is pooled P times. The P1 pooled input is the output of the C2 convolutional layer, the j vector Conv in the j core C2jPooling is carried out, the step length is set to be s, and the calculation formula is as follows:
Poolj＝[max(Convj1,…,Convjp)…max(Convj(1+s)…Convj(1+s)+p)…]
the output of the maximum pooling layer is sequentially input into two convolution layers, then passes through a global maximum pooling layer and a full-connection layer, and finally the predicted value is output
The hyper-parameter setting training Batch Batch Size of the one-dimensional convolutional neural network is 10, the training round epochs is 100, an early stopping mode is adopted, the overfitting is prevented, and the algorithm convergence is not good.
Calculating a cross entropy loss function L according to the output of the full link layer, and calculating the error between the real value and the predicted value of the sample, wherein the formula is as follows:
wherein N in the formula represents the total number of samples,
calculating an updating gradient according to a cross entropy loss function L, and then updating network parameters, wherein an Adam optimization algorithm is adopted as the optimization algorithm, and the method specifically comprises the following steps:
mt＝μ*mt-1+(1-μ)*gt
in the above formula, gtIs the gradient of the objective function, mt，ntRepresenting the first and second moment estimates of the gradient, and μ, v are the exponential decay rates of the first and second moments of the gradient, respectively. At the time of the t-th iteration,
the one-dimensional convolutional neural network parameters are updated as follows:
in the equation, α represents a learning rate and is an infinitesimal number.
And updating parameters of the neural network according to the formula, performing verification test by using a 10-fold cross verification mode, starting the safety event classification model trained in the current round if the effect on the verification set is poor, and otherwise, outputting the safety event classification model trained in the current round until complete iteration. The training process visualization is shown in fig. 4.
Finally, a target security event classification model is determined from a plurality of security event classification models obtained through a plurality of rounds of training, and the method can be specifically realized through the following steps:
after the obtained framing signals are labeled with corresponding classification labels, extracting all the framing signals according to a second preset proportion to form an acoustic signal test sample set;
performing performance evaluation on each safety event classification model based on the acoustic signal test sample set;
and taking the security event classification model with the optimal performance as a target security event classification model.
After the acoustic signal training sample set is determined in the above manner, the rest is the acoustic signal test sample set. And inputting the prepared acoustic signal test sample set into each trained safety event classification model, and outputting the classification result of the test. For the comparison of the tested classification result and the actual label, the performance of various measures such as Accuracy, Precision, Recall, F-score and the like is calculated, and the detailed formula is as follows:
in the equation, TP represents the number of samples predicted to be positive and actually matched, FP represents the number of samples predicted to be positive but actually negative, FN represents the number of samples predicted to be negative but actually positive, and TN represents the number of samples predicted to be negative but actually negative. Accuracy refers to the ratio of predicted to actual correct to the total number of all samples. The Precision rate Precision refers to the number of actual positive examples in all the samples of the positive examples, and the Recall rate Recall refers to the number of samples of the positive examples which are predicted to be correct, which is the number of samples of the positive examples. And F-score is a comprehensive evaluation index of accuracy and recall rate.
And finally, calculating a confusion matrix according to the formula, and analyzing the classification condition of each type of the model. As shown in fig. 5. And confirming a model with the optimal performance evaluation as a target security event classification model.
Finally, the acoustic signals detected by the acoustic sensor can be identified through a target safety event classification model,
specifically, the central control device sorts data of each acquisition node into 5s data frames, then inputs each data frame into a target security event classification model for classification prediction, wherein a plurality of events are predicted finally, the data events are stored and automatically labeled with classification labels of the security events, and then algorithm model iterative training is performed after manual examination of the labels, so that the model can identify the intrusion events more and more accurately.
And finally, carrying out safety event classification and identification on signals acquired by acoustic sensors in different security areas through a target safety event classification model, and controlling an alarm device to alarm according to a corresponding alarm strategy under the condition that the safety event with the target type is identified. For example, if a wall-turning type security event is identified, an alarm device may be triggered to alarm, the alarm device may be a speaker device, or may also be an information transmission device that notifies monitoring personnel of alarm information, and the like.
According to the method in the embodiment, through the designed one-dimensional convolutional neural network structure, the distributed acoustic vibration sensor in the security area is used for collecting the sample sets of various security events and is used for learning and training the algorithm, the algorithm is used for carrying out automatic feature mining on the sensor vibration signals, the events can be accurately and rapidly identified, and the final result is superior to that of the traditional methods such as manual feature extraction and time-frequency domain conversion. The method in the embodiment is different from other CNN methods in that the sampling frequency of the signal is different, and the high-efficiency extraction of the low-frequency signal can realize real-time accurate monitoring in a wider range; secondly, the network structures are different, and the network structure which is designed autonomously is adopted to achieve a better recognition effect.
Referring to fig. 6, a second embodiment of the present invention further provides a central control device, which is applied to a distributed acoustic sensing security system, where the central control device is connected to acoustic sensors disposed in different security areas, and the central control device includes:
the determining unit 601 is configured to determine an acoustic signal training sample set, where the acoustic signal training sample set includes acoustic signals corresponding to M types of security events, each acoustic signal is labeled with a classification tag of the security event, each acoustic signal is acquired by a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1;
a model training unit 602, configured to perform multiple rounds of training on the signals in the training sample set through a one-dimensional convolutional neural network that is set up in advance, where the one-dimensional convolutional neural network includes a preliminary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer, and a Softmax layer for M classification, and in each round of training, the signals in the training sample set are input to the preliminary feature extraction layer for primary feature extraction, the preliminarily extracted features are compressed through the maximum pooling layer, and then input to the deep feature extraction layer for deep feature extraction, and then after the compressed features are averaged and pooled through the global average pooling layer, the full connection layer is input for fusion, and finally the Softmax layer is used for classification to obtain a classification result, and a loss value of the classification result and an actual classification label is calculated through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training;
and the security unit 603 is configured to perform security event classification and identification on signals acquired by the acoustic sensors in different security areas through the target security event classification model, and control the alarm device to alarm according to a corresponding alarm strategy when a security event of a target type is identified.
As an optional embodiment, the determining unit is specifically configured to:
acquiring acoustic signals corresponding to various event security events of a first preset duration acquired by acoustic sensors corresponding to different time periods;
and according to a sliding window with second preset time duration, performing framing processing on each obtained acoustic signal in a mode of sliding for third preset time duration each time, labeling corresponding classification labels on the obtained framing signals, and then extracting all framing signals according to a first preset proportion to form the acoustic signal training sample set.
As an optional embodiment, the preliminary feature extraction layer includes a first convolution sublayer and a second convolution sublayer that are sequentially connected, the deep feature extraction layer includes a third convolution sublayer and a fourth convolution sublayer that are sequentially connected, a Dropout layer is further disposed between the full connection layer and the Softmax layer of the one-dimensional convolutional neural network, and the features are discarded through the Dropout layer.
As an alternative embodiment, the model training unit is specifically configured to:
and performing verification test by adopting a 10-fold cross verification mode, starting and stopping early if the classification precision on the verification sample set is smaller than the preset precision to obtain the safety event classification model of the training of the current round, and otherwise, outputting the safety event classification model of the training of the current round until complete iteration.
As an alternative embodiment, the model training unit is specifically configured to:
after the obtained framing signals are labeled with corresponding classification labels, extracting all the framing signals according to a second preset proportion to form an acoustic signal test sample set;
performing performance evaluation on each safety event classification model based on the acoustic signal test sample set;
and taking the security event classification model with the optimal performance as a target security event classification model.
The specific implementation of the safety protection performed by the central control device in this embodiment has been described in detail in the foregoing first embodiment, and reference may be made to the contents in the first embodiment, which is not described herein again.
Referring to fig. 7, for convenience of description, only the parts related to the embodiment of the present invention are shown, and details of the specific technology are not disclosed, please refer to the method part of the embodiment of the present invention.
Fig. 7 is a schematic diagram illustrating a partial structure of a central control device according to an embodiment of the present invention. The central control device includes a memory 701, and the memory 701 is used for storing a program for executing the safety protection method in the foregoing first embodiment. The central control device further comprises a processor 702 connected to the memory 701, wherein the processor 702 is configured to execute the program stored in the memory 701. The central control equipment is connected with acoustic sensors arranged in different security areas.
The processor 702, when executing the computer program, implements the steps of the security protection method in the first embodiment. Alternatively, the processor implements the functions of the modules/units in the central control device in the second embodiment when executing the computer program.
Illustratively, the computer program may be partitioned into one or more modules/units that are stored in the memory and executed by the processor to implement the invention. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution of the computer program in the computer apparatus.
The device may include, but is not limited to, a processor, a memory. It will be understood by those skilled in the art that the schematic diagram 7 is merely an exemplary diagram of functional components of the central control device, and does not constitute a limitation of the central control device, and may include more or less components than those shown, or combine some components, or different components, for example, the central control device may further include an input-output device, a network access device, a bus, and the like.
The Processor 702 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like which is the control center for the computer device and which connects the various parts of the overall computer device using various interfaces and lines.
The memory 701 may be used for storing the computer programs and/or modules, and the processor may implement various functions of the computer apparatus by operating or executing the computer programs and/or modules stored in the memory and calling data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, application programs (such as a sound playing function, an image playing function, etc.) required by one or more functions, and the like; the storage data area may store data (such as audio data, video data, etc.) created according to the use of the central control apparatus, etc. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), one or more magnetic disk storage devices, a Flash memory device, or other volatile solid state storage devices.
In an embodiment of the present invention, the processor 702 has the following functions:
determining an acoustic signal training sample set, wherein the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired through a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1;
performing multi-round training on signals in the training sample set through a one-dimensional convolutional neural network which is set up in advance, wherein the one-dimensional convolutional neural network comprises a primary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer and an M-class Softmax layer which are arranged in sequence, in each round of training, inputting the signals in the training sample set into the primary feature extraction layer for primary feature extraction, compressing the primary extracted features through the maximum pooling layer, inputting the compressed features into the deep feature extraction layer for deep feature extraction, performing average pooling on the compressed features through the global average pooling layer, inputting the full connection layer for fusion, finally classifying through the Softmax layer to obtain a classification result, and calculating the loss value of the classification result and an actual classification label through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training;
and carrying out classified identification on the safety events of the signals acquired by the acoustic sensors in different security areas through the target safety event classification model, and controlling an alarm device to alarm according to a corresponding alarm strategy under the condition that the safety event with the target type is identified.
In an embodiment of the present invention, the processor 702 has the following functions:
acquiring acoustic signals corresponding to various event security events of a first preset duration acquired by acoustic sensors corresponding to different time periods;
and according to a sliding window with second preset time duration, performing framing processing on each obtained acoustic signal in a mode of sliding for third preset time duration each time, labeling corresponding classification labels on the obtained framing signals, and then extracting all framing signals according to a first preset proportion to form the acoustic signal training sample set.
The primary feature extraction layer comprises a first convolution sublayer and a second convolution sublayer which are connected in sequence, the deep feature extraction layer comprises a third convolution sublayer and a fourth convolution sublayer which are connected in sequence, a Dropout layer is further arranged between a full connection layer and a Softmax layer of the one-dimensional convolution neural network, and the Dropout layer discards the features.
In an embodiment of the present invention, the processor 702 has the following functions:
and performing verification test by adopting a 10-fold cross verification mode, starting and stopping early if the classification precision on the verification sample set is smaller than the preset precision to obtain the safety event classification model of the training of the current round, and otherwise, outputting the safety event classification model of the training of the current round until complete iteration.
In an embodiment of the present invention, the processor 702 has the following functions:
after the obtained framing signals are labeled with corresponding classification labels, extracting all the framing signals according to a second preset proportion to form an acoustic signal test sample set;
performing performance evaluation on each safety event classification model based on the acoustic signal test sample set;
and taking the security event classification model with the optimal performance as a target security event classification model.
A fourth embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored, and the central control apparatus integrated functional unit in the second embodiment of the present invention may be stored in one computer-readable storage medium if it is implemented in the form of a software functional unit and sold or used as a stand-alone product. Based on such understanding, all or part of the processes in the security protection method according to the first embodiment may be implemented by a computer program, which may be stored in a computer-readable storage medium and used by a processor to implement the steps of the above-described method embodiments. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying said computer program code, medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, etc. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.
A fifth embodiment of the present invention provides a distributed acoustic sensing security system, including the central control device in the third embodiment and acoustic sensors connected to the central control device and disposed in different security areas, where each acoustic sensor of the central control device acquires an acoustic signal corresponding to a corresponding security area and transmits the acoustic signal to the central control device, and the central control device is configured to perform security protection by using the method in the first embodiment.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A safety protection method is characterized in that the method is applied to central control equipment of a distributed acoustic sensing security system, the central control equipment is connected with acoustic sensors arranged in different security areas, and the method comprises the following steps:
determining an acoustic signal training sample set, wherein the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired through a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1;
performing multi-round training on signals in the training sample set through a one-dimensional convolutional neural network which is set up in advance, wherein the one-dimensional convolutional neural network comprises a primary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer and an M-class Softmax layer which are arranged in sequence, in each round of training, inputting the signals in the training sample set into the primary feature extraction layer for primary feature extraction, compressing the primary extracted features through the maximum pooling layer, inputting the compressed features into the deep feature extraction layer for deep feature extraction, performing average pooling on the compressed features through the global average pooling layer, inputting the full connection layer for fusion, finally classifying through the Softmax layer to obtain a classification result, and calculating the loss value of the classification result and an actual classification label through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training;
and carrying out classified identification on the safety events of the signals acquired by the acoustic sensors in different security areas through the target safety event classification model, and controlling an alarm device to alarm according to a corresponding alarm strategy under the condition that the safety event with the target type is identified.
2. The method of claim 1, wherein the determining a set of acoustic signal training samples comprises:
acquiring acoustic signals corresponding to various event security events of a first preset duration acquired by acoustic sensors corresponding to different time periods;
and according to a sliding window with second preset time duration, performing framing processing on each obtained acoustic signal in a mode of sliding for third preset time duration each time, labeling corresponding classification labels on the obtained framing signals, and then extracting all framing signals according to a first preset proportion to form the acoustic signal training sample set.
3. The method of claim 1, wherein the preliminary feature extraction layer comprises a first convolution sublayer and a second convolution sublayer which are connected in sequence, the deep feature extraction layer comprises a third convolution sublayer and a fourth convolution sublayer which are connected in sequence, a Dropout layer is further arranged between a full-connection layer and a Softmax layer of the one-dimensional convolutional neural network, and features are discarded through the Dropout layer.
4. The method of claim 1, wherein verifying the model for updating network parameters by verifying signals in the sample set until a convergence condition is reached comprises:
and performing verification test by adopting a 10-fold cross verification mode, starting and stopping early if the classification precision on the verification sample set is smaller than the preset precision to obtain the safety event classification model of the training of the current round, and otherwise, outputting the safety event classification model of the training of the current round until complete iteration.
5. The method of claim 2, wherein determining the target security event classification model from a plurality of security event classification models obtained from a plurality of rounds of training comprises:
after the obtained framing signals are labeled with corresponding classification labels, extracting all the framing signals according to a second preset proportion to form an acoustic signal test sample set;
performing performance evaluation on each safety event classification model based on the acoustic signal test sample set;
and taking the security event classification model with the optimal performance as a target security event classification model.
6. The utility model provides a well accuse equipment, its characterized in that is applied to distributed acoustic sensing security protection system, well accuse equipment is connected with the acoustic sensor who sets up in different security protection regions, well accuse equipment includes:
the device comprises a determining unit, a judging unit and a processing unit, wherein the determining unit is used for determining an acoustic signal training sample set, the acoustic signal training sample set comprises acoustic signals corresponding to M types of security events, each acoustic signal is marked with a classification label of the security event, each acoustic signal is acquired through a corresponding acoustic sensor in the distributed acoustic sensing detection system, and M is an integer greater than 1;
the model training unit is used for performing multi-round training on the signals in the training sample set through a pre-built one-dimensional convolutional neural network, wherein the one-dimensional convolutional neural network comprises a primary feature extraction layer, a maximum pooling layer, a deep feature extraction layer, a global average pooling layer, a full connection layer and an M-classified Softmax layer which are sequentially arranged, in each round of training, the signals in the training sample set are input into the primary feature extraction layer for primary feature extraction, the primary extracted features are compressed through the maximum pooling layer and then input into the deep feature extraction layer for deep feature extraction, the compressed features are averaged and pooled through the global average pooling layer and then input into the full connection layer for fusion, finally the Softmax layer is used for classification to obtain a classification result, and the loss value of the classification result and an actual classification label is calculated through a loss function, updating the network parameters of the one-dimensional convolutional neural network model by using a back propagation algorithm, verifying the model for updating the network parameters by verifying signals in a sample set until a convergence condition is reached to obtain a security event classification model for each round of training, and determining a target security event classification model from a plurality of security event classification models obtained by a plurality of rounds of training;
and the security unit is used for carrying out security event classification and identification on signals acquired by the acoustic sensors in different security areas through the target security event classification model, and controlling the alarm device to alarm according to a corresponding alarm strategy under the condition of identifying the existence of a security event of a target type.
7. The device according to claim 6, wherein the determining unit is specifically configured to:
acquiring acoustic signals corresponding to various event security events of a first preset duration acquired by acoustic sensors corresponding to different time periods;
and according to a sliding window with second preset time duration, performing framing processing on each obtained acoustic signal in a mode of sliding for third preset time duration each time, labeling corresponding classification labels on the obtained framing signals, and then extracting all framing signals according to a first preset proportion to form the acoustic signal training sample set.
8. The central control device is applied to a distributed acoustic sensing security system and comprises a processor and a memory:
the memory for storing a program for performing the method of any one of claims 1-5;
the processor is configured to execute programs stored in the memory.
9. A computer storage medium storing computer software instructions for a method of safeguarding according to any of claims 1 to 5, comprising a program designed to carry out the above aspects.
10. A distributed acoustic sensing security system is characterized by comprising a central control device and acoustic sensors which are connected with the central control device and are arranged in different security areas, wherein after each acoustic sensor of the central control device collects an acoustic signal corresponding to a corresponding security area, the acoustic signal is transmitted to the central control device, and the central control device is used for carrying out security protection by the method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010168130.1A CN111476102A (en) | 2020-03-11 | 2020-03-11 | Safety protection method, central control equipment and computer storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010168130.1A CN111476102A (en) | 2020-03-11 | 2020-03-11 | Safety protection method, central control equipment and computer storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111476102A true CN111476102A (en) | 2020-07-31 |
Family
ID=71748313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010168130.1A Pending CN111476102A (en) | 2020-03-11 | 2020-03-11 | Safety protection method, central control equipment and computer storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111476102A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112329535A (en) * | 2020-09-29 | 2021-02-05 | 国网四川省电力公司经济技术研究院 | CNN-based quick identification method for low-frequency oscillation modal characteristics of power system |
| CN113286311A (en) * | 2021-04-29 | 2021-08-20 | 沈阳工业大学 | Distributed perimeter security protection environment sensing system based on multi-sensor fusion |
| CN116506483A (en) * | 2023-06-30 | 2023-07-28 | 中科海慧（北京）科技有限公司 | Intelligent park building security method based on wireless transmission |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107545890A (en) * | 2017-08-31 | 2018-01-05 | 桂林电子科技大学 | A kind of sound event recognition method |
| CN108932480A (en) * | 2018-06-08 | 2018-12-04 | 电子科技大学 | The study of distributing optical fiber sensing signal characteristic and classification method based on 1D-CNN |
| CN109473120A (en) * | 2018-11-14 | 2019-03-15 | 辽宁工程技术大学 | A kind of abnormal sound signal recognition method based on convolutional neural networks |
-
2020
- 2020-03-11 CN CN202010168130.1A patent/CN111476102A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107545890A (en) * | 2017-08-31 | 2018-01-05 | 桂林电子科技大学 | A kind of sound event recognition method |
| CN108932480A (en) * | 2018-06-08 | 2018-12-04 | 电子科技大学 | The study of distributing optical fiber sensing signal characteristic and classification method based on 1D-CNN |
| CN109473120A (en) * | 2018-11-14 | 2019-03-15 | 辽宁工程技术大学 | A kind of abnormal sound signal recognition method based on convolutional neural networks |
Non-Patent Citations (4)
| Title |
|---|
| WU H J ET AL: "One-dimensional CNN-based intelligent recognition of vibrations in pipeline monitoring with DAS", 《LIGHT TECHNOL》 * |
| 左冬森: "光纤周界安防入侵振动信号识别与定位方法研究", 《中国优秀硕士学位论文全文数据库》 * |
| 张建同: "《实用多元统计分析》", 31 August 2016 * |
| 朱频频: "《智能客户服务技术与应用》", 31 March 2019 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112329535A (en) * | 2020-09-29 | 2021-02-05 | 国网四川省电力公司经济技术研究院 | CNN-based quick identification method for low-frequency oscillation modal characteristics of power system |
| CN112329535B (en) * | 2020-09-29 | 2023-03-24 | 国网四川省电力公司经济技术研究院 | CNN-based quick identification method for low-frequency oscillation modal characteristics of power system |
| CN113286311A (en) * | 2021-04-29 | 2021-08-20 | 沈阳工业大学 | Distributed perimeter security protection environment sensing system based on multi-sensor fusion |
| CN113286311B (en) * | 2021-04-29 | 2024-04-12 | 沈阳工业大学 | Distributed perimeter security environment sensing system based on multi-sensor fusion |
| CN116506483A (en) * | 2023-06-30 | 2023-07-28 | 中科海慧（北京）科技有限公司 | Intelligent park building security method based on wireless transmission |
| CN116506483B (en) * | 2023-06-30 | 2023-10-03 | 中科海慧（北京）科技有限公司 | Intelligent park building security method based on wireless transmission |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111027378B (en) | Pedestrian re-identification method, device, terminal and storage medium | |
| CN111476102A (en) | Safety protection method, central control equipment and computer storage medium | |
| CN109473119B (en) | Acoustic target event monitoring method | |
| WO2022105336A1 (en) | Image classification method and electronic device | |
| CN111738044A (en) | Campus violence assessment method based on deep learning behavior recognition | |
| CN116546023B (en) | Method and system for identifying violent behaviors of oil and gas operation area | |
| CN111859010A (en) | Semi-supervised audio event identification method based on depth mutual information maximization | |
| CN114926825A (en) | Vehicle driving behavior detection method based on space-time feature fusion | |
| CN115100739B (en) | Man-machine behavior detection method, system, terminal device and storage medium | |
| CN114023354A (en) | Guidance type acoustic event detection model training method based on focusing loss function | |
| CN115830399A (en) | Classification model training method, apparatus, device, storage medium, and program product | |
| CN115272656A (en) | Environment detection alarm method and device, computer equipment and storage medium | |
| CN113707175B (en) | Acoustic event detection system based on feature decomposition classifier and adaptive post-processing | |
| CN114266271A (en) | Distributed optical fiber vibration signal mode classification method and system based on neural network | |
| Xiao et al. | AMResNet: An automatic recognition model of bird sounds in real environment | |
| Pandeya et al. | Sound event detection in cowshed using synthetic data and convolutional neural network | |
| CN113780094A (en) | Perimeter intrusion identification method based on Gaussian mixture model and hidden Markov model | |
| CN116311081B (en) | Medical laboratory monitoring image analysis method and system based on image recognition | |
| CN110123484B (en) | Livestock delivery detection method and device | |
| CN116933002A (en) | Action detection method and device, terminal equipment and storage medium | |
| CN114140879A (en) | Behavior identification method and device based on multi-head cascade attention network and time convolution network | |
| CN112232329A (en) | Multi-core SVM training and alarming method, device and system for intrusion signal recognition | |
| CN113571092A (en) | Method for identifying abnormal sound of engine and related equipment thereof | |
| CN117830313B (en) | Method, device and system for detecting lower part of oilfield wellhead based on deep learning | |
| CN115936431B (en) | Re-crime risk assessment method, device, computer equipment and readable storage medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200731 |
|
| RJ01 | Rejection of invention patent application after publication |