60 results about "Chaotic neural network" patented technology

An inventory forecasting model and its construction method based on chaotic neural network. The inventory of finished products is the key factor in precise distribution. If the inventory of finished products is sufficient, accurate delivery will be guaranteed, but the high inventory of finished products will bring a negative impact on the enterprise. The risk is high. On the one hand, it is difficult to process other materials after the original roll is processed into finished products. Once the user does not use it, it is likely to become a waste product. On the other hand, the finished product inventory takes up a large inventory space, which will make Limited storage capacity is getting tighter. The present invention divides the work into two phases. The first is the learning phase. The data of all the distribution users of the sample companies in the past three years are used as samples to establish a model, and these samples are used to learn and adjust the connection weight coefficients of the chaotic neural network, so that the network Realize the given input-output relationship; then the implementation stage, use the trained neural network to obtain the expected effect, establish a perfect calculation model, and realize the reasonable setting of the inventory.

The invention relates to a chaotic neural network encryption communication circuit, in particular to an encryption communication circuit based on a chaotic neuron network with a time delay state. A clear text signal i(t) of a transmission end drives a first time delay chaotic neural network system through a reversed phase amplification circuit, the first time delay chaotic neural network system outputs a chaotic signal x(t), the chaotic signal x(t) and the clear text signal i(t) are overlapped to generate a signal x(t)+i(t), an encryption transmission signal s(t) is generated through an encryption scheme circuit; and the encryption transmission signal s(t) is transmitted to a receiving end through a transmission channel, the signal x(t)+i(t) is solved through corresponding decryption scheme circuit to drive a second time delay chaotic neural network system, the second time delay chaotic neural network system generates a corresponding chaotic signal y(t) synchronous with the chaotic signal x(t), and the signal x(t)+i(t) is subtracted from the chaotic signal y(t) to obtain a clear text signal r(t). The chaotic neural network encryption communication circuit overcomes the defects that confidentiality of a common low-dimensional chaotic system is poor and a high-dimensional chaotic system is difficult to be physically implemented, implements encryption transmission function of theclear text signal, and effectively simplifies a real circuit device.

The invention discloses an optimal allocation method of basin water resources based on a multi-objective chaotic genetic algorithm, and belongs to the technical field of optimal allocation of water resources. The method includes the steps of obtaining basic information of basin water resources; establishing a multi-objective water resource optimal allocation mathematical model and performing allocation model parameter calibration; solving a water resource optimal allocation alternative scheme set by using a multi-objective chaotic genetic algorithm; and finally, determining a best equilibrium scheme of water resource optimal allocation through a chaotic neutral network comprehensive evaluation model. According to the invention, the chaotic ergodicity and the inversion of the genetic algorithm are coupled, the speed of the algorithm is improved, the optimal solution is stable, and the multi-objective optimal allocation requirements of a basin water resource complex system are met.

The invention discloses an intelligent diagnosis system for a power supply and a protection method. Mainly, a failure of the power supply system is judged by the following two manners: 1) judging whether the power supply system works normally by using an infrared signal spectrum and a sound spectrum, 2) finding all the potential failure modes through a failure model and impact analysis based on apower supply circuit model, performing simulation operation under all the failure modes and obtaining simulation data so that the simulation data is more targeted and representative; and constructingthe power supply failure model in a manner of training a chaotic neural network through the simulation data. The power supply failure information obtained through the two manners not only contains thepower supply failure type but also contains an integrated chip or an electronic element, which breaks down in the power supply circuit, thereby greatly facilitating the subsequent inspection and maintenance of the worker, reducing the workload, and improving the working efficiency.

The invention provides an analysis method for infant brain medical computer scanning images and a realization system. The analysis method is to process the brain medical computer scanning images into images with prominent fractal characteristics, and then perform mathematic quantitative analysis on the images by a chaos fractal analysis method. The method is realized under Matlab R2007 through programming, utilizes chaotic neural network and fractal principles, performs serialized division, identification and fractal mode analysis on the infant brain medical computer scanning images to obtain fractal dimension quantified reference values and quantified reference values of multifractal spectrum width of normal infant brain medical computer scanning images in different age brackets, realizes clinical prediction on sick infants with intelligent disability and brain paralysis which have no typical neural image manifest characteristics by taking the reference values as a comparison standard, thoroughly changes the conventional analysis mode for the infant brain medical computer scanning images, and has high accuracy and strong repeatability compared with scoring results of the clinically widely used Gesell scale.

The invention discloses a radar multi-target tracking optimization method based on the chaotic neural network. The method is mainly characterized in that state one-step prediction of a tth target at the k time, measurement prediction of the tth target at the k time, the measurement prediction information of a j'th measurement for the tth target at the k time, a one-step prediction error covariance matrix of the tth target at the k time, an information covariance matrix of the tth target at the k time, Kalman gain of the tth target at the k time, an nk*T'-dimension measurement-target association matrix at the k time, an (nk+1)*T'-dimension effective likelihood function matrix of association of nk meausurements and T' targets at the k time, an (nk+1)*T'-dimension normalization matrix of association of the nk meausurements and the T' targets at the k time, an (nk+1)*T'-dimension accurate probability matrix of association of the nk meausurements and the T' targets at the k time, a state equation of the tth target at the k time and an error covariance matrix of the tth target at the k time are sequentially calculated; the t is respectively made to be 1 to T', the error covariance matrix of the T'th targets at the k time is acquired, and real-time tracking for the T'th targets is carried out by a radar according to the error covariance matrix of the T'th targets at the k time.

The present invention proposes a nonlinear self-feedback chaotic neural network signal blind detection method, which uses a nonlinear function as the self-feedback item of the chaotic neural network, and applies the double Sigmoid function to the blind detection method. In each iteration, it first enters the chaos neural network, and then into the second activation function. Because the chaotic neural network has the advantage of being able to avoid being trapped in a local optimum, the present invention inherits this characteristic of the chaotic neural network and improves blind detection performance; and, compared with the chaotic neural network of the linear self-feedback item, the non-linear self-feedback The chaotic neural network has more complex dynamic behavior, which makes the internal state of the network have more efficient chaotic search ability and search efficiency. Under the same conditions, the method of the invention has better anti-noise performance than the traditional Hopfield signal blind detection method.

The invention provides a signal blind detection method based on a double Sigmoid chaotic neural network. According to the method, by means of the chaotic neural network and a second activation function, the double Sigmoid chaotic neural network is formed, each time iteration is carried out, the chaotic neural network is logged in firstly and then the second activation function is logged in. Due to the fact that the chaotic neural network has the advantage of being capable of avoiding being stuck in the local minimum, blind detection performance is improved, anti-noise performance of the network operation speed is improved, and the method is superior to a traditional Hopfield signal blind detection algorithm.

The invention discloses a secret communication method based on a time-lag memristor chaos neural network. According to the method, a driving system and a response system are established by utilizing a two-dimensional time-lag memristor chaos neural network and a simple synchronous controller is designed to enable a clear text signal to be coded to achieve a good secret communication effect in transmission. The method overcomes the defects that a conventional chaos neural network is poor in secrecy performance and fixed in weight and consumes much network energy and the like, and provides a brand-new solution to secret communication and transmission of signals.

The invention discloses a face recognition method and device under a side face condition, equipment and a storage medium, and the method comprises the steps: obtaining local binary features corresponding to face feature points in a received face image, carrying out the regression of the local binary features, and recognizing the shape of a face; when the recognized face shape is a non-front face,reconstructing the face image through a pre-constructed front face reconstruction model to generate a front face image; according to the front face image, extracting a face frame through a pre-constructed transient chaotic neural network; according to the face frame, extracting a face feature vector through a pre-constructed FaceNet network model; splicing the face feature vectors, and calculatingthe similarity between the spliced face feature vectors and the front face image sample; obtaining a front face image sample corresponding to the maximum similarity, and outputting the front face image sample as a face recognition result. According to the method, the accuracy of face recognition can be effectively improved under the condition of side face or partial occlusion.

The invention discloses a river flow monitoring data quality control method based on a chaotic neural network. The method comprises the following steps: a, sorting historical flow data in a chronological order to obtain time series data; b, normalizing the time series data; c, calculating the optimal time lag [tau] and the optimal embedding dimension m, and converting the one-dimensional time series data into multi-dimensional spatial sample data; d, proportionally dividing the multi-dimensional spatial sample data into multi-dimensional spatial training sample data and multi-dimensional spatial test sample data; e, using the multi-dimensional spatial training sample data to train and construct a GMDH neural network, using the multi-dimensional spatial test sample data to test the GMDH neural network so as to obtain a GMDH neural network model; f, detecting the abnormal value of river flow monitoring data; g, inspecting the completeness of the data within 24 hours of a day; and h, storing the revised river flow monitoring data is stored in a database. The method is able to improve the quality of river flow monitoring data.

The invention belongs to the field of machine vision and target identification, and in particular relates to an optimized calculation-based characteristic point matching method. The method comprises the following steps of: respectively detecting characteristic points in a template image and a target image by using window area direction variation as a measured value of the characteristic points; controlling the number of the characteristic points by setting a characteristic threshold value; determining a matching criterion energy function according to the relative position information and gray scale information of the characteristic points; and performing optimized calculation on the energy function by using a hysteretic and chaotic neural network to solve a matching result of the characteristic points in the two images. The method can be applied to a target identification system.

The invention discloses a neuron oscillator, namely a Lee-oscillator. The Lee-oscillator comprises excitation neurons and inhibitory neurons. The excitation neurons are used for receiving inhibitory signals from the inhibitory neurons. The inhibitory neurons are used for receiving excitation signals from the excitation neurons. Excitation self-feedback respectively exists in the excitation neurons and the inhibitory neurons. The Lee-oscillator further comprises input neurons and output neurons. The output neurons are used for receiving excitation signals from the input neurons, the excitation signals from the excitation neurons and the inhibitory signals from the inhibitory neurons. The input neurons and the excitation neurons are respectively used for receiving stimulation of external input. The Lee-oscillator has continuity of neural dynamics and provides actual gradual changing from chaotic dynamics to non-chaotic dynamics. Additionally provided is an instant chaotic self-associative network based on the Lee-oscillator.

The invention provides a signal blind detection method based on a complex sinusoidal chaotic neural network. A non-monotone excitation function combined by a complex sinusoidal chaotic neural network and Sigmoid, time-varying output function gains and a piecewise exponential annealing function are adopted to form a complex sinusoidal chaotic neural network; in case of each iteration, a chaotic neural network is entered firstly, and an activation function is then entered; being trapped in the local minimum can be prevented by the chaotic neural network. The method of the invention inherits the features of the chaotic neural network, and the blind detection performance is improved; the network has richer and more flexible transient chaos dynamics characteristics and stronger global searching capability; and in the same condition, the anti-noise performance is better than that of the traditional Hopfield signal blind detection algorithm.

The invention discloses a secret communication method of a parameter uncertain time-delay chaotic neural network, and belongs to the technical field of secret communication. According to the secret communication method of the parameter uncertain time-delay chaotic neural network, firstly, a driving system is built, then a response system is built according to the driving system, and then an anti-synchronization controller is built according to the driving system and the response system. When the ciphertext signal is transmitted, the driving system generates a chaotic signal, superposes the chaotic signal and the ciphertext signal to obtain a superposed signal, and then transmits the superposed signal to the response system through a channel. And the response system generates an anti-synchronization chaotic signal through an anti-synchronization controller, and obtains a decrypted ciphertext signal according to the superposed signal and the anti-synchronization chaotic signal. The secret communication method of the time-delay chaotic neural network overcomes the defect that an existing chaotic neural network secret communication technology is weak in anti-interference capability, the parameters of the secret communication method of the time-delay chaotic neural network are uncertain, and the anti-interference capability of secret communication is improved.

The invention relates to a watermark encryption and decryption algorithm based on a chaotic neural network, and belongs to the technical field of information hiding. The characteristic that Logistic chaotic mapping is extremely sensitive to an initial value and the good approximation effect of the Chebyshev neural network are utilized, and an LCNN network is established to encrypt the watermark, so that the security of the watermark is ensured. The network weights obtained by the LCNN neural network during each operation are different, and the secret keys for watermark encryption are also different, so that the encryption algorithm obtained based on the LCNN has a large secret key space, and the security is high. Besides, when the chaos initial value is slightly changed, the secret key isgreatly changed, the encrypted watermarks obtained after encryption are different from one another, the expected goal of one-time pad can be achieved, and the safety of the algorithm is improved.

The invention relates to a dynamic frequency allocation method for a cognitive radio cellular network, belonging to the technical field of cognitive radio. The dynamic frequency allocation method comprises the following steps: creating a real-time interference frequency list according to available frequency spectrum information of each cell sensed by a base station and a cognitive mobile subscriber; then creating a cost function according to the number of frequencies required by each cell and inter-cell interference constraint conditions; and minimizing the cost function by a noise chaotic neural network method, thus meeting the frequency requirement of each cell, avoiding interference to an authorized user, also avoiding inter-cell interference and maximizing the frequency spectrum utilization. Meanwhile, in the invention, the frequency with minimum number of each cell is locked, so that each cell can be relatively stably allocated with one frequency as a common control channel.

The invention discloses a time-lag neural network hyperchaos circuit which belongs to the field of nonlinear circuits. The circuit comprises an integration circuit, a time lag module and an excitation module, and can generate a chaotic signal under the interference of a pulse signal. The integration circuit consists of an operational amplifier U1, an operational amplifier U2, an operational amplifier U5, an operational amplifier U6, an operational amplifier U3, and an operational amplifier U9. The time lag module comprises a first time lag module HB1 and a second time lag module HB5. The excitation module comprises a first excitation module HB2, a second excitation module HB3, a third excitation module HB4, and a fourth excitation module HB6. Compared with the prior art, the circuit has the improvement that a time-lag hyperchaos neural network phase diagram is represented under different interferences through adjusting a pulse voltage value.

The invention discloses a heterogeneous sensor network encryption protocol based on a chaotic neural network public key encryption algorithm; the public key encryption algorithm based on the chaotic neural network is combined with characteristics of a heterogeneous sensor network, thus designing a novel encryption protocol suitable for the heterogeneous sensor network; the encryption protocol specifically comprises the following steps: building a base station layer and a cluster head layer secrecy communication network; building a cluster head layer and a perception layer secrecy communication network; updating a cluster head layer secret key; updating a perception layer secret key; adding a novel cluster head node; adding a novel perception node; quitting the cluster head node; and quitting the perception node. The encryption protocol can flexibly apply a public key code system into the sensor network, thus solving the problems that when a sensing node storage room and the calculation ability are different in the heterogeneous sensor network, only the secret keys with the same secret key room size can be fixedly used to encode data; the heterogeneous sensor network encryption protocol can improve network safety and encryption flexibility.

The invention relates to a method for secret communication of a chaotic neural network under a signal quantization circumstance. The method comprises the following steps of (1) establishing a chaotic neural network model and a quantizer model; (2) constructing a state feedback controller to obtain an error dynamics system; (3) solving a controller gain matrix K and substituting into an actual controller to obtain a synchronous controller; (4) loading a ciphertext signal through a driving system in order to obtain a superposed signal and transmitting the superposed signal to a response system through a network; (5) making the driving system and the response system synchronous under the action of the synchronous controller; and (6) obtaining a recovered ciphertext signal through the superposed signal and a synchronizing signal. The method considers the uniform quantization phenomenon in the network environment and provides the synchronous controller, the driving system and the response system are synchronous under the action of the synchronous controller, and the recovered ciphertext signal is obtained through the superposed signal and the synchronizing signal after quantization, so that the impact of uniform quantization and random disturbance can be effectively eliminated and the secret communication can be carried out under the chaotic neural network model.

The invention discloses a signal blind detection method based on a double Sigmoid hysteresis noise chaotic neural network, which is characterized by comprising the following steps: step SS1: constructing a received data matrix XN; step SS2: performing singular value decomposition on the received data matrix XN; Step SS3: setting a weight matrix W; step SS4: selecting an activation function of thedouble Sigmoid hysteresis chaotic neural network, performing the iterative operation of the double Sigmoid hysteresis chaotic neural network, and then substituting the result iterated each time into an energy function E (t) of the double Sigmoid hysteresis noise chaotic neural network; when the energy function E (t) reaches a minimum value, the double Sigmoid hysteresis noise chaotic neural network reaching equilibrium, and ending the iteration. The present invention utilizes a double Sigmoid chaotic neural network and hysteresis noise for the first time to form a double Sigmoid hysteresis noise chaotic neural network, which enhances the network optimization performance and improves the quality of the network optimization solution. The anti-noise performance and the convergence speed of the method in present invention are superior to those of the conventional Hopfield signal blind detection algorithm.

The invention discloses a hyperchaos neural network hiding secret communication circuit which is belongs to the technical field of communication. The circuit comprises a drive circuit and a response circuit, which are respectively a hyperchaos neural network circuit. A plaintext signal u(t) is superposed with a hyperchaos signal x(t) generated by the drive circuit, thereby generating a superposed signal x(t)+u(t). A password signal m(t) is obtained through encryption based on the conventional cryptography, is transmitted to the response signal through a transmission channel, and then is decrypted, thereby obtaining a superposed signal x(t)+u(t). The superposes signal is used for driving the response circuit, thereby obtaining a corresponding synchronization signal y(t). Finally, the superposed signal x(t)+u(t) is subtracted by y(t), thereby obtaining a password signal v(t). Compared with a conventional circuit, a hiding neural network is a hyperchaos neural network, thereby enabling a system to be more unpredictable, and improving the communication security.

The invention discloses a blind detection method of a noise chaotic neural network based on discrete multilevel hysteresis. The method includes the following steps: constructing a received data matrixXN; performing singular value decomposition on the received data matrix XN; setting a weight matrix WRI, and constructing a performance function; introducing a piecewise annealing function into the chaotic neural network to construct a discrete multilevel hysteretic chaotic neural network based on piecewise annealing; constructing a dynamic equation of an improved new model of the noise chaotic neural network based on discrete multilevel hysteresis, performing iterative operation on the dynamic equation of the improved new model, then substituting the result of each iteration into an energy function E(t) of the noise chaotic neural network based on discrete multilevel hysteresis, and when the energy function E(t) reaches a minimum value, determining that the chaotic neural network based on discrete multilevel hysteresis reaches a balance and the iteration ends. According to the scheme of the invention, an activation function is improved, a noise chaotic neural network model based on discrete multilevel hysteresis is constructed, and the phenomenon that the neural network falls into a minimal value point can be better avoided.

The invention relates to a hybrid encryption method for controlling system network security and a system thereof, and the method comprises the steps: generating a first group of initial values corresponding to a password through first equipment, and generating a second group of initial values corresponding to to-be-encrypted data; wherein the first group of initial values encrypts a password through a chaotic neural network, and the second group of initial values encrypts data through the chaotic neural network; and respectively encrypting the first group of initial values and the second groupof initial values through SM2, and sending all encrypted data to the second device.

The invention provides a security multicast communication method based on a chaotic neural network. The security multicast communication method is characterized by comprising the steps of combining a symmetric cryptographic algorithm based on the chaotic neural network and a centralized key management protocol LKH, and applying the combination to multicast communication, wherein according to the chaotic neural network, the chaos theory is introduced into a neural network to enable an artificial neural network to have abundant non-linear dynamic characteristics and high computational complexity, and chaotic characteristics can be achieved through multiple neural network models, such as a disperse Hopfield neural network having the chaotic characteristics after improvement; according to the symmetric cryptographic algorithm based on the chaotic neural network, output of concurrent LFSRs is taken as input of the disperse Hopfield neural network having the chaotic characteristics, and random selection is performed on pseudorandom sequences generated by the LFSRs by means of the non-linear dynamic characteristics and the chaotic characteristics of the neural network. Therefore, in a secret key of the symmetric algorithm (as specified in the specification), T0 is a connection weight singular matrix, H is a circulant matrix, and H' is a transposed matrix of H.

The embodiment of the invention provides a network device, a method for network synchronization, a communication method, and devices. The network device comprises N circuit units which are parallel, wherein each circuit unit comprises neurons and at least two memristors; m memristors in the at least two memristors are used for being connected with a real-time neuron; n memristors in the at least two memristors are used for being connected with a time lag neuron; each neuron comprises an RC oscillating circuit, a real-time signal processing unit and a time lag signal processing unit; one end of the RC oscillating circuit is grounded, and the other end is connected with the at least two memristors, the real-time signal processing unit and the time lag signal processing unit; the real-time signal processing unit is used for determining excitation of a real-time neuron state according to the real-time neuron state; and the time lag signal processing unit is used for determining the excitation of a time delay neuron state according to the time delay neuron state. According to the embodiment of the network device, the method for network synchronization, the communication method, and the devices, the memristors are taken as the synaptic connection of the neurons, a chaos neural network is simple in structure, and the expandability is high.

The invention relates to a radar cooperative information sharing distribution path optimization method based on a chaotic neural network. The method is mainly suitable for the radar cooperative information sharing distribution path optimization in a radar cooperative process of a centerless node. The main flow is as follows: firstly, constructing a target optimization function with constraint conditions according to the basic requirements of a radar cooperative information sharing distribution path optimization problem; then constructing an energy function of the radar cooperative information sharing distribution path optimization problem; constructing a chaotic neuron model; constructing an exponential function self-feedback chaotic neural network model; and finally using the exponential function self-feedback chaotic neural network to perform radar cooperative information sharing distribution path optimization. The method provided by the invention has the advantages of simple implementation method, good optimization effect and sufficient theoretical foundation of used method, and the radar cooperative information sharing distribution path optimization problem in radar cooperative detection of the centerless node can be solved quickly and effectively.

The invention provides a dynamic channel allocation method facing the OFDMA cellular network, and belongs the technical field of mobile communication. The method is realized by the steps that (1) thecellular network is colored in three colors, and the channel demand numbers of three colors of cells which are adjacent to one another and provided with the maximal sum of the channel demand numbers is obtained; (2) according the channel demand numbers of the three colors of cells, continuous OFDMA channels from the number 1 are generated; (2) first OFDMA channel allocation is carried out by taking the generated OFDMA channels as the cellular network; and (3) if the first OFDA channel allocation cannot satisfy the demand for the channel numbers of all cells, second OFDMA channel allocation iscarried out via a retarded noise chaotic neural network. Twice OFDMA channel allocation is sued to reduce the total required OFDMA channel number and the computing amount of dynamic channel allocation, and the instantaneity of radio mobile communication is ensured.