A large animal intrusion detection system and method using trees

Abstract

The present invention discloses a large animal intrusion detection system and method using trees. It includes a detection unit and a signal processor. The detection unit is arranged on the tree trunk and is connected to the signal processor; each detection unit includes a metal ring and an electromagnetic transceiver. A plurality of metal rings are wound around the tree trunk and arranged at intervals, and the metal ring at the bottom of the tree trunk is electrically connected to the electromagnetic transceiver as a coupling metal ring; the detection units on each tree are controlled to work by a polling mechanism, and the S21 parameter is obtained by the detection unit. The S21 parameter is obtained in advance by placing large animals in different positions and input into the BP neural network model for training; in the case of detection, the S21 parameter is obtained by polling and processing by the detection unit, and the position of the large animal is judged by analysis and processing. The present invention can achieve comprehensive coverage detection based on trees, and the equipment used has low cost, can protect wild animals, and has good application prospects in aspects such as forestry production safety.

Description

Technical Field

[0001] The present invention relates to a method for locating animal targets in the wild forest, and particularly to a large animal intrusion detection system and method using trees. Background Art

[0002] The existing large wild animal systems include a variety of high-performance sensing sensors. The corresponding traditional positioning methods are visual recognition and ultrasonic and radar detection. However, in the current technology: Visual recognition technology cannot work in the complex wild environment with poor light conditions or obstacles. In addition, it needs to process a large number of details of target images, and the image processing algorithm is very complex, with very high requirements for the MCU and signal processor. Finally, the extraction and positioning of scene features will be very difficult and cannot guarantee good real-time performance; Microwave and ultrasonic radar have a small detection range, and full-coverage work requires a large amount of manpower and material resources, resulting in waste. Summary of the Invention

[0003] Aiming at the deficiencies in the background art, the design purpose of the present invention is to propose a large animal intrusion detection system and method using trees.

[0004] To achieve the above purpose, the technical solution adopted by the present invention is:

[0005] I. A large animal intrusion detection system using trees:

[0006] It includes a detection unit and a signal processor; the detection unit is arranged on the trunk of each tree, and the detection units on all trees are connected to the signal processor; each detection unit includes a metal ring and an electromagnetic transceiver, the metal ring is wound around the trunk, and the metal ring and the electromagnetic transceiver are electrically connected.

[0007] Each detection unit includes a plurality of metal rings, and the plurality of metal rings are arranged at vertical intervals along the trunk direction. One of the metal rings at the bottom of the trunk is electrically connected to the electromagnetic transceiver as a coupling metal ring.

[0008] The metal ring is composed of multiple turns of metal coils wound around the trunk.

[0009] The electromagnetic transceiver includes a polling arbiter, a power amplifier, a software radio module, and a direct digital synthesis frequency synthesizer DDS; both ends of the coupling metal ring are connected to the polling arbiter, the polling arbiter is connected to the direct digital synthesis frequency synthesizer DDS through the power amplifier, the polling arbiter is connected to the signal processor through the software radio module, and at the same time, the signal processor and the polling arbiter are directly connected.

[0010] The polling arbiter controls the direct digital synthesis frequency synthesizer DDS and the software radio module to switch between working states, and they work respectively at the transmission time and the reception time:

[0011] At the emission time, the direct digital synthesizer frequency DDS emits a broadband signal with continuously changing frequency, which is sent to the coupling metal ring through a power amplifier to excite the coupling metal ring. The coupling metal ring excites a high-frequency current distribution on multiple other metal rings on the tree, and then forms a non-radiative high-frequency electromagnetic near-field distribution around the tree.

[0012] At the reception time, a current is induced and received by the coupling metal ring from the high-frequency electromagnetic near-field distribution, and after passing through the polling arbiter and the software radio module, the current is collected and processed to obtain the S21 parameter, and then the S21 parameter is sent to the signal processor.

[0013] The number of the detection units, the number of electromagnetic transceivers is the same as the number of tree trunks where the coupling metal rings are placed.

[0014] Multiple detection units work in a polling manner. Each time during polling, the coupling metal ring of one tree serves as the transmitting end, and the coupling metal rings of all other trees serve as the receiving ends.

[0015] In the present invention, multiple detection units are deployed on multiple adjacent trees in a tree area. When a large animal passes through the tree area, the electromagnetic signals received by one or more detection units change due to the proximity of the large animal, and then the position of the invading large animal is judged according to the change.

[0016] II. A method for detecting large animal intrusion using trees, the method comprising the following steps:

[0017] 1) Arrange detection units on the trees in the tree area, control the detection units on each tree to work by a polling mechanism, and obtain the S21 parameter by the detection unit;

[0018] 2) Establish an xy two-dimensional coordinate system for the plane of the tree area. Place large animals at different positions in advance and process them according to the method in step 1) to obtain the S21 parameter, and input all the S21 parameters into the BP neural network model for training;

[0019] 3) In the case of detection to be performed, the detection unit performs polling processing to obtain the S21 parameter, and analyzes and processes all the S21 parameters to judge the position where the large animal is located.

[0020] 8. A method for detecting large animal intrusion using trees according to claim 7, wherein: the specific content of step 1) is:

[0021] In each detection unit, the direct digital synthesizer frequency DDS and the software radio module are controlled by the polling arbiter to switch between working modes, so that the detection unit serves as the transmitting end or the receiving end;

[0022] Under each polling, one of the detection units of the trees is controlled as the transmitting end, and the detection units of the remaining trees are used as the receiving ends to work. The software radio module in each detection unit acting as the receiving end receives the current and processes it to obtain the S21 parameter.

[0023] Poll the detection units of different trees as the transmitting end. The polling mechanism is controlled by the signal processor to perform high-frequency switching work.

[0024] The input of the BP neural network model is the S21 parameters of different polls; the output of the BP neural network model is the regression values of different positions.

[0025] The specific content of step 3) is as follows:

[0026] 3.1) In the case of detection to be performed, all the S21 parameters obtained by each polling of the detection unit are input into the trained BP neural network regression model for processing to output the regression values of different positions. For the regression value of each position, the probability of the presence of large animals at each position is obtained by processing according to the following formula:

[0027]

[0028] Among them, P(x, y) represents the probability that large animals exist at the coordinate position (x, y), Z(x, y) represents the regression value output by the BP neural network model at the coordinate position (x, y), p represents the maximum coordinate along the x direction, q represents the maximum coordinate along the y direction, and e represents the natural logarithm;

[0029] 3.2) Repeat the operation of step 3.1) for M polls. Place one probability density distribution obtained from each poll in a row of the observation mixing matrix W. Each probability in each probability density distribution is placed in the same row of the observation mixing matrix W after being sorted by number, so as to reconstruct the M collected probability density distributions into the following observation mixing matrix W:

[0030]

[0031] Among them, P i (x, y) represents the probability that large animals exist at the coordinate position (x, y) under the Mth poll, i = 1, 2,..., M; one probability density distribution obtained from each poll is located in a row of the observation mixing matrix W, and each probability in the probability density distribution is located in the same row of the observation mixing matrix W after being sorted by number;

[0032] 3.3) Calculate the probability distribution of large animals at different positions according to the observation mixing matrix W:

[0033]

[0034] K(index) = max(Q)

[0035] Wherein, Q is the product of the probability density distributions of M polling times, W(j) represents the j-th row of the observation mixing matrix W, j represents the row ordinal number of the observation mixing matrix W, and max() represents taking the maximum element in the matrix; K(index) is the maximum probability value of the position where the large animal is located, and index is the column number value corresponding to the maximum probability value in the observation mixing matrix W;

[0036] The position where the large animal is located is calculated according to the column number value index according to the following formula:

[0037]

[0038]

[0039] Wherein, x and y respectively represent the coordinate positions where the large animal is located.

[0040] The large animal mentioned refers to large wild animals, such as elephants, bears, and tigers.

[0041] The system of the present invention places a metal ring at the lower end of the tree, places multiple metal rings in the middle of the tree, and installs a transceiver for transmitting and receiving electromagnetic signals on the coupled metal rings. The transceiver transmits a signal with continuously changing frequency in the transmitting mode and excites a high-frequency current distribution on multiple metal rings of the tree through an excitation structure, thereby forming a non-radiating high-frequency electromagnetic near-field distribution around the tree. The transceiver receives the high-frequency electromagnetic near-field signals of adjacent trees through the tree and the metal rings in the receiving mode.

[0042] The method of the present invention deploys multiple transceivers on multiple adjacent trees. When a large animal passes through the tree area, the electromagnetic signals received by one or more receivers change due to the proximity of the animal, and then the type, position, and quantity of the invading animal are judged according to the change.

[0043] The beneficial effects of the present invention are:

[0044] The present invention provides a brand-new sensing solution for locating large animals in the wild and complex environment. Compared with the current vision and radar technologies, it can reduce the influence of the external environment, increase the detection coverage range, and save costs.

[0045] The present invention controls the working mode of the electromagnetic transceiver by adopting a polling arbitration method. When a large animal passes through the tree area, the near-field electromagnetic signals received by the receiver change due to the proximity of the animal, and then the type, position, and quantity of the invading animal are judged according to the change. The system and method have good application prospects in protecting wild animals, agricultural production safety, etc.

[0046] Moreover, the system and method of the present invention can protect wild animals and has good application prospects in aspects such as forestry production safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of the device setup in the woods in an embodiment of the present invention;

[0048] Figure 2 It is a block diagram of the electromagnetic transceiver in an embodiment of the present invention;

[0049] Figure 3 It is the animal position coordinate grid in an embodiment of the present invention;

[0050] Figure 4 It is a schematic diagram of the neural network model of the signal processor.

[0051] In the figure: 1. Electromagnetic transceiver, 2. Coupling metal ring, 3. Software radio module, 4. Polling arbiter, 5. Power amplifier, 6. Signal processor, 7. Direct digital synthesis frequency synthesizer DDS. DETAILED DESCRIPTION OF THE INVENTION

[0052] The present invention will be further described below with reference to the accompanying drawings.

[0053] As Figure 1 shown, the system includes a detection unit and a signal processor 6; the method is for detecting and positioning large animal targets around trees. A detection unit is arranged on the trunk of each tree, and all the detection units on the trees are connected to the signal processor 6; each detection unit includes a metal ring and an electromagnetic transceiver 1. The metal ring is composed of multiple turns of metal coils wound around the trunk. The metal ring is wound around the trunk and is electrically connected to the electromagnetic transceiver 1.

[0054] Each detection unit includes multiple metal rings, and the multiple metal rings are arranged at a certain vertical interval along the trunk direction. One of the metal rings at the bottom of the trunk is electrically connected to the electromagnetic transceiver 1 as the coupling metal ring 2, and the remaining metal rings are not electrically connected to the electromagnetic transceiver. Except for the coupling metal ring 2 connected to the electromagnetic transceiver, which is an open-loop metal coil, the remaining metal rings are all closed-loop metal coils.

[0055] As Figure 2 shown, the electromagnetic transceiver 1 includes a polling arbiter 4, a power amplifier 5, a software radio module 3, and a direct digital synthesis frequency synthesizer DDS; both ends of the coupling metal ring 2 are electrically connected to the polling arbiter 4. The polling arbiter 4 is connected to the direct digital synthesis frequency synthesizer DDS through the power amplifier 5, and the polling arbiter 4 is connected to the signal processor 6 through the software radio module 3. At the same time, the signal processor 6 is directly connected to the polling arbiter 4.

[0056] The signal processor 6 directly sends a control signal to the polling arbiter 4 to control the polling arbiter 4 to switch on the power amplifier 5 or the software radio module 3.

[0057] When the software radio module 3 is switched on, the polling arbiter 4 sends the induced current received by the coupled metal ring 2 to the signal processor 6 via the software radio module 3.

[0058] The direct digital synthesis frequency synthesizer DDS and the software radio module 3 are respectively used for transmitting signals and receiving signals, and the polling arbiter 4 is responsible for switching between transmitting signals and receiving signals.

[0059] The polling arbiter 4 controls the direct digital synthesis frequency synthesizer DDS and the software radio module 3 to switch to work, and they work respectively at the transmitting time and the receiving time:

[0060] At the transmitting time, the direct digital synthesis frequency synthesizer DDS emits a broadband signal with continuously changing frequency, which is sent to the coupled metal ring 2 via the power amplifier and excites the coupled metal ring 2. The coupled metal ring 2 excites a high-frequency current distribution on multiple other metal rings on the tree, and then forms a non-radiating high-frequency electromagnetic near-field distribution around the tree;

[0061] At the receiving time, and the coupled metal ring 2 induces and receives the high-frequency electromagnetic near-field distribution to generate a current. After passing through the polling arbiter 4 and the software radio module 3, the current is collected and processed to obtain the S21 parameter, and then the S21 parameter is sent to the signal processor 6.

[0062] That is, the signal processor 6 is externally connected to the electromagnetic transceiver 1 and controls the working mode of the current electromagnetic transceiver 1 through the polling arbiter 4, which is the transmitting mode or the receiving mode. In the transmitting mode, a continuous frequency-changing signal in a certain frequency range is used to excite the metal ring near the tree trunk by the DDS7 and the power amplifier 5. The input of the power signal amplifier is connected to the square-wave output pin of the DDS, and the output of the power signal amplification module is connected to the polling arbiter. In the receiving mode, the software radio 3 transmits the collected signal feature S21 parameter to the signal processor 6.

[0063] The input end of the power signal amplification module is connected to the square-wave output pin of the direct digital synthesis frequency synthesizer DDS, and the output end of the power signal amplification module is connected to the polling arbiter. The direct digital synthesis frequency synthesizer DDS outputs a square-wave signal. The signal processor changes the frequency of the square-wave signal output by changing the frequency division ratio. The power signal amplification module amplifies the square-wave signal, and the square-wave signal output by the power signal amplification module is used as the excitation signal of the radiator.

[0064] In specific implementation, the chip of the direct digital synthesizer DDS7 is AD9851, which is a highly integrated product produced by Analog Devices, Inc. of the United States using advanced DDS direct digital frequency synthesis technology. The chip of the power signal amplifier 5 is TPA3116D2, which is a class-D stereo amplifier with at least 90% operating efficiency and with AM interference prevention function. DDS7 outputs a square wave signal, the signal processor 5 changes the frequency of the square wave signal output by changing the frequency division ratio, and the power signal amplification module amplifies the square wave signal, and the square wave signal output by the power signal amplifier 6 is used as the transmission signal of the coupling metal ring.

[0065] The polling arbiter is a kind of radio frequency switch, and the signal processor gives a control signal to determine the working mode of the electromagnetic transceiver.

[0066] Multiple detection units work in a polling manner. Each time of polling, the coupling metal ring of one tree serves as the transmitting end, and the coupling metal rings of all the other trees will serve as the receiving ends.

[0067] The relationship between the total number of polling times M and the number of detection units N is expressed as:

[0068]

[0069] In specific implementation, N = 4 and M = 6.

[0070] The specific implementation working process of the present invention is as follows:

[0071] 1) Detection units are arranged on the trees in the tree area, and the detection units on each tree are controlled to work by the polling mechanism, and the S21 parameter is obtained by detection by the detection unit;

[0072] Within each detection unit, the direct digital synthesizer DDS and the software radio module 3 are controlled by the polling arbiter 4 to switch work, so that the detection unit serves as the transmitting end or the receiving end; when the direct digital synthesizer DDS works, the detection unit serves as the transmitting end; when the software radio module 3 works, the detection unit serves as the receiving end.

[0073] Under each polling, one of the detection units of the trees is controlled to serve as the transmitting end, and the detection units of the other trees serve as the receiving ends to work. The software radio module 3 in each detection unit serving as the receiving end receives and processes the current to obtain the S21 parameter and sends it to the signal processor 6.

[0074] 2) An xy two-dimensional coordinate system is established for the plane of the tree area. Large animals are pre-placed at different positions and processed in the manner of step 1) to obtain the S21 parameter, and all the S21 parameters are input into the BP neural network model for training.

[0075] AsFigure 3 As shown in the figure, the input of the BP neural network model is the S21 parameters of different polls. Different input neurons correspond to the S21 parameters obtained from different polls, and the same input neuron corresponds to the S21 parameters obtained from the same poll.

[0076] The output of the BP neural network model is the regression values at different positions. The probability distribution of the presence of large animals at different positions is calculated based on the regression values at different positions, and the probability of the presence of large animals at each position is calculated based on the regression value of each position. Different output neurons correspond to the regression values at different positions, and the same output neuron corresponds to the regression value at the same position.

[0077] As Figure 4 shown, the BP neural network model is specifically defined; as Figure 3 According to the grid division, the number of nodes in the input layer of the neural network is 128, the number of nodes in the first hidden layer is selected as 64, the number of nodes in the second hidden layer H2 is 32, and the number of nodes in the output layer is 8×8 = 64.

[0078] The defined loss is as follows:

[0079]

[0080] where m is the size of the data set, is the label value of a certain data in the data set, is the predicted value of the BP neural network positioning algorithm.

[0081] During the training process, the neuron states are passed layer by layer in progression. If the error between the final predicted value and the actual value is greater than the error tolerance, the error is propagated back to the previous layers, and the weights of the model are corrected accordingly until the change in loss in 10 iterations is less than 0.1, then the training stops.

[0082] 3) In the case of detection to be performed, the polling unit performs polling processing to obtain the S21 parameters, and analyzes and processes all the S21 parameters to determine the position where the large animal is located.

[0083] 3.1) In the case of detection to be performed, all the S21 parameters obtained by each polling of the detection unit are input into the trained BP neural network regression model to process and output the regression values at different positions. For the regression value of each position, the following formula is used to process to obtain the probability of the presence of large animals at each position:

[0084]

[0085] Among them, P(x, y) represents the probability that a large animal exists at the coordinate position (x, y) in the xy two-dimensional coordinate system, Z(x, y) represents the regression value output by the BP neural network model at the coordinate position (x, y) in the xy two-dimensional coordinate system, p represents the maximum coordinate of the xy two-dimensional coordinate system in the x direction, q represents the maximum coordinate of the xy two-dimensional coordinate system in the y direction, e represents the natural logarithm, (x, y) represents the grid coordinates within the animal detection coverage range, and the grid size is 50 cm × 50 cm.

[0086] 3.2) Repeat the operation in step 3.1) for M times of polling. Place each probability density distribution obtained from each polling in a row of the observation mixing matrix W. Each probability in each probability density distribution is placed in the same row of the observation mixing matrix W after being sorted by number. Thus, the M collected probability density distributions are reconstructed into the following observation mixing matrix W:

[0087]

[0088] Among them, P i (x, y) represents the probability that a large animal exists at the coordinate position (x, y) in the xy two-dimensional coordinate system under the Mth polling, i = 1, 2,..., M; each probability density distribution obtained from each polling is located in a row of the observation mixing matrix W, and each probability in the probability density distribution is located in the same row of the observation mixing matrix W after being sorted by number;

[0089] Among them, the size of the observation mixing matrix W is 6 rows and 8×8 columns.

[0090] 3.3) Calculate the probability distribution of the large animal at different positions according to the observation mixing matrix W:

[0091]

[0092] K(index) = max(Q)

[0093] Among them, Q is the product of the probability density distributions of M times of polling, W(j) represents the jth row of the observation mixing matrix W, j represents the row ordinal number of the observation mixing matrix W, max() represents taking the maximum element in the matrix; K(index) is the maximum probability value of the position where the large animal is located. Finally, the coordinate value of the animal within the detection range is obtained, index is the column number value corresponding to the maximum probability value in the observation mixing matrix W;

[0094] Calculate the position where the large animal is located according to the column number value index according to the following formula:

[0095]

[0096]

[0097] Among them, x and y respectively represent the coordinate positions in the xy two-dimensional coordinate system where the large animal is located.

Claims

1. A method for detecting large animal intrusion using trees, characterized in that: The method uses a large animal intrusion detection system using trees. The system includes a detection unit and a signal processor (6); the detection unit is arranged on the trunk of each tree, and the detection units on all trees are connected to the signal processor (6); each detection unit includes a metal ring and an electromagnetic transceiver (1), the metal ring is wound around the trunk, and the metal ring is electrically connected to the electromagnetic transceiver (1); The method includes the following steps: 1) Arrange detection units on the trees in the tree area, and control the detection units on each tree to work by a polling mechanism, and obtain the S21 parameter by the detection unit; 2) Establish an xy two-dimensional coordinate system for the tree area, pre-place large animals at different positions and process them in the manner of step 1) to obtain the S21 parameter, and input all the S21 parameters into the BP neural network model for training; 3) In the case of detection to be performed, the detection unit performs polling processing to obtain the S21 parameter, and analyzes and processes all the S21 parameters to determine the position where the large animal is located; The specific content of step 3) is: 3.1) In the case of detection to be performed, input all the S21 parameters obtained by each polling of the detection unit into the trained BP neural network regression model for processing and output the regression values at different positions. Then, process the regression value at each position according to the following formula to obtain the probability of the presence of a large animal at each position: where P(x,y) represents the probability that a large animal exists at the coordinate position (x,y), Z(x,y) represents the regression value output by the BP neural network model at the coordinate position (x,y), p represents the maximum coordinate along the x direction, q represents the maximum coordinate along the y direction, and e represents the natural logarithm; 3.2) Repeat the operation of step 3.1) for M times of polling, place each probability density distribution obtained by each polling in a row of the observation mixing matrix W, and place each probability in each probability density distribution in the same row of the observation mixing matrix W after sorting according to the number, so as to reconstruct the M collected probability density distributions into the following observation mixing matrix W: Among them, P i (x, y) represents the probability that a large animal exists at the coordinate position (x, y) in the M-th polling, where i = 1, 2, …, M; a probability density distribution obtained in each polling is located in a row of the observation mixing matrix W, and each probability in the probability density distribution is located in the same row of the observation mixing matrix W after being sorted by number. 3.3) Calculate the probability distribution of large animals at different positions according to the observation mixing matrix W: K(index) = max(Q) where Q is the product of the probability density distributions of M times of polling, W(j) represents the j-th row of the observation mixing matrix W, j represents the row ordinal number of the observation mixing matrix W, and max() represents taking the maximum element in the matrix; K(index) is the maximum probability value of the position where the large animal is located, and index is the column number value corresponding to the maximum probability value in the observation mixing matrix W; Calculate the position where the large animal is located according to the column number value index according to the following formula: where x and y respectively represent the coordinate positions where the large animal is located.

2. The method for detecting large animal intrusion using trees according to claim 1, characterized in that: The specific content of step 1) is: Within each detection unit, the direct digital synthesis frequency synthesizer DDS and the software radio module (3) are controlled by a polling arbiter (4) to switch between operating modes, enabling the detection unit to function as either a transmitter or a receiver. During each polling cycle, one detection unit of a tree is controlled to act as the transmitter, while the detection units of the remaining trees act as receivers. The software radio module (3) within each detection unit acting as the receiver receives the current and processes it to obtain the S21 parameter.

3. A method for detecting large animal intrusion using trees according to claim 1, characterized in that: The input of the BP neural network model is the S21 parameters from different polling cycles; the output of the BP neural network model is the regression values at different positions.

4. A method for detecting large animal intrusion using trees according to claim 1, characterized in that: Each detection unit includes a plurality of metal rings, which are arranged at vertical intervals along the trunk direction. One metal ring at the bottom of the trunk is electrically connected to the electromagnetic transceiver (1) as the coupling metal ring (2).

5. A method for detecting large animal intrusion using trees according to claim 1, characterized in that: The metal ring is composed of multiple turns of metal wire wound around the trunk.

6. A method for detecting large animal intrusion using trees according to claim 4, characterized in that: The electromagnetic transceiver (1) includes a polling arbiter (4), a power amplifier (5), a software radio module (3), and a direct digital synthesis frequency synthesizer DDS; both ends of the coupling metal ring (2) are connected to the polling arbiter (4), the polling arbiter (4) is connected to the direct digital synthesis frequency synthesizer DDS via the power amplifier (5), the polling arbiter (4) is connected to the signal processor (6) via the software radio module (3), and at the same time, the signal processor (6) is directly connected to the polling arbiter (4).

7. A method for detecting large animal intrusion using trees according to claim 4, characterized in that: The polling arbiter (4) controls the direct digital synthesis frequency synthesizer DDS and the software radio module (3) to switch between operating modes, each operating during the transmission time and the reception time respectively: During the transmission time, the direct digital synthesis frequency synthesizer DDS emits a broadband signal with continuously varying frequency, which is sent to the coupling metal ring (2) via the power amplifier and excites the coupling metal ring (2). The coupling metal ring (2) induces a high-frequency current distribution on the other multiple metal rings on the tree, thereby forming a non-radiative high-frequency electromagnetic near-field distribution around the tree. During the reception time, the coupling metal ring (2) inductively receives the high-frequency electromagnetic near-field distribution to generate a current, which is collected and processed by the polling arbiter (4) and the software radio module (3) to obtain the S21 parameter, and then the S21 parameter is sent to the signal processor (6).

8. A method for detecting large animal intrusion using trees according to claim 4, characterized in that: Multiple detection units operate in a polling manner. During each polling cycle, the coupling metal ring of one tree serves as the transmitter, and the coupling metal rings of all the other trees serve as receivers.

Images