Multi-view modeling data fusion device

Abstract

The invention discloses a multi-view modeling data fusion device, and relates to the technical field of data processing, and the device comprises the following components: a quantum entanglement module, data coding and modulation, quantum communication and synchronization, and fusion algorithm and processing. Through combination of quantum entanglement, quantum communication and an advanced fusion algorithm, efficient and accurate fusion of multi-view data is realized, the device can realize nearly real-time data transmission and synchronization by utilizing the over-distance effect of quantum entanglement, the timeliness of data fusion is remarkably improved, and meanwhile, the real-time performance of the device is improved. By introducing a quantum error correction mechanism and an advanced fusion algorithm, the device can effectively reduce the error rate in the data transmission process, the accuracy and reliability of a fusion result are ensured, and powerful technical support is provided for constructing a more comprehensive and accurate multi-view model.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and specifically provides a multi-view modeling data fusion device. Background Art

[0002] With the rapid development of information technology, multi-view data fusion technology has been widely applied in various fields, such as intelligent monitoring, medical diagnosis, and autonomous driving. These application scenarios often require processing data from multiple perspectives or sensors simultaneously to obtain more comprehensive and accurate information. Through integrating and analyzing data from different perspectives, multi-view data fusion technology can significantly improve the system's perception ability, decision-making ability, and execution efficiency.

[0003] Traditional technologies have deficiencies. On the one hand, traditional data transmission methods are limited by the physical transmission speed and it is difficult to achieve real-time data synchronization, which is particularly disadvantageous in application scenarios that require rapid response. On the other hand, during data transmission, due to noise and interference factors, traditional methods are prone to bit errors, affecting the accuracy and reliability of data fusion. Secondly, traditional fusion algorithms are often based on fixed rules and models and are difficult to adapt to complex and changing data environments and application requirements, resulting in limited fusion effects.

[0004] In summary, traditional multi-view data fusion technologies have many deficiencies when dealing with large-scale and high-dimensional data and are difficult to meet the requirements of modern applications for efficient and accurate data fusion. Therefore, it is particularly important to develop a multi-view modeling data fusion device. Summary of the Invention

[0005] The purpose of the present invention is to make up for the deficiencies of the prior art and provide a multi-view modeling data fusion device. It can achieve efficient and accurate fusion of multi-view data by introducing quantum entanglement, quantum communication, and advanced fusion algorithms, providing strong technical support for constructing a more comprehensive and accurate multi-view model.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A multi-view modeling data fusion device, which includes the following components: a quantum entanglement module, data encoding and modulation, quantum communication and synchronization, and a fusion algorithm and processing.

[0007] The quantum entanglement module: A quantum entanglement generation and control module is integrated inside each sub-device. This module can generate pairs of quantum bits in an entangled state and allocate a part of the quantum bits to other sub-devices that need to work together. By precisely controlling the state of the quantum bits, the accuracy and stability of data during transmission are ensured.

[0008] The data encoding and modulation: At the transmitting end, the sub-device encodes the collected multi-view data and converts it into an information form that can be carried by quantum bits. Using quantum modulation technology, the data information is loaded onto the quantum bits in the entangled state, so that they carry multi-view data features;

[0009] The quantum communication and synchronization: when a sub-device needs to share multi-view data with other sub-devices, by operating the local entangled quantum bits, according to the long-distance effect of quantum entanglement, the quantum bits of the remote sub-device entangled with it will instantly undergo corresponding state changes. After the remote sub-device receives the quantum bits with changed states, it restores the original multi-view data through decoding operations, thereby realizing rapid data synchronization;

[0010] The fusion algorithm and processing: After receiving the multi-perspective data synchronized from other sub-devices, each sub-device uses a fusion algorithm to fuse the local data and the remote synchronized data. The algorithm fully considers the characteristics and relevance of the multi-perspective data, and can complete data fusion in a short time to build a more comprehensive and accurate multi-perspective model.

[0011] Furthermore, the data acquisition module of the sub-device includes multiple types of sensor interfaces. For different types of sensor data, the sub-device adopts an adaptive sampling algorithm for data acquisition. The algorithm is: assuming that the sensor data is a time series x(t), the sampling frequency f s The adjustment is based on the formula where f s0 is the initial sampling frequency, α is the frequency adjustment coefficient, which is determined by the relationship between historical data fluctuations and model accuracy and obtained through multiple experimental fittings, n is the number of data points in the sampling window, is the mean of the data in the window, σ is the standard deviation of the data in the window. For data with sharp fluctuations, the sampling frequency is increased to obtain more accurate data. For stable data, the sampling frequency is reduced to reduce data redundancy and improve the efficiency and quality of overall data collection.

[0012] Furthermore, the quantum entanglement generation and control module generates entangled quantum bit pairs through an improved spontaneous parametric down-conversion technology based on nonlinear optical processes. Specifically, an additional phase modulator is introduced, and its modulation parameter φ is given by the formula Determine, where P in is the input laser power, P thLet \(P_0\) be the threshold power and \(\beta\) be the modulation coefficient, which is determined according to the optimization results of the entanglement state fidelity in the experiment. By adjusting this phase modulator, the generation efficiency and quality of the entangled state qubit pairs can be effectively improved. The fidelity of the generated entangled state qubit pairs is increased by \(\gamma\%\) compared with the traditional method, where \(\gamma\) is obtained through experimental measurement, ensuring the accuracy and stability of data during quantum communication transmission and providing a more reliable basis for subsequent data synchronization.

[0013] Furthermore, in the data encoding stage, a dynamic encoding algorithm based on the superposition state of qubits is adopted. For numerical data in multi-view data, i.e., sensor measurement values, the data range is divided into \(m\) intervals, and each interval corresponds to a qubit superposition state. Let the data value be \(d\), and its corresponding qubit superposition state \(|\psi\rangle\) is determined by the formula where \(a\) i is the superposition coefficient, \(c\) i is the central value of the \(i\)-th interval, and \(\sigma\) i is the width correlation coefficient of the \(i\)-th interval, which is determined according to the discreteness of the data distribution. For text data, it is first converted into word vectors through natural language processing, and then the numerical values of each dimension of the word vectors are encoded according to the above numerical data encoding method. This dynamic encoding algorithm can flexibly adjust the encoding method according to the actual distribution of the data, improve the encoding efficiency and accuracy, and enhance the data carrying capacity on qubits.

[0014] Furthermore, during the quantum communication and synchronization process, to ensure the reliability of data transmission, a quantum error correction mechanism is introduced. When an error in the qubit state is detected, an error correction algorithm based on quantum entanglement redundancy is adopted. Let the original entangled state qubit pair sent by the sender be \(|\psi\) AB \rangle\), and at the same time, an additional pair of auxiliary entangled state qubit pairs \(|\psi\) CD \rangle\) is sent. After the receiver receives the qubits, the qubit state information is obtained through measurement operations. If an error is detected, the error amount is calculated according to the formula \(\Delta = |\langle\psi' AB |\psi AB \rangle - \langle\psi' CD |\psi CD \rangle|\), where \(|\psi AB '\) and \(|\psi CD '\) are the qubit states actually received by the receiver. Then, according to the pre-established error correction table, which is generated through a large number of simulation experiments based on different error types and corresponding correction operations, the incorrect qubit state is corrected. Through this error correction mechanism, the bit error rate in the quantum communication process can be effectively reduced, ensuring the accuracy of multi-view data during the synchronization process.

[0015] Furthermore, in the data fusion stage, for the fusion of image data in multi-view data, an algorithm based on multi-scale feature fusion and attention mechanism is adopted. Let the input multi-view images be I1, I2, …, I n , first perform multi-scale decomposition on each image to obtain feature maps F ij at different scales, where i represents the image number and j represents the scale level. Then calculate the attention weight w ij of each feature map. The formula is where g is the feature extraction function, is the k-th feature channel of the feature map F ij , θ k is the feature channel weight, which is determined by analyzing the correlation between different image features and the final fusion effect. Finally, the fused image feature F fusion is This algorithm can highlight important image features, improve the quality of image data fusion, and provide more accurate image information for subsequent multi-view model construction.

[0016] Furthermore, in the data fusion stage, for the fusion of time series data in multi-view data, a dynamic fusion algorithm based on adaptive weights is adopted. Let multiple time series data be S1(t), S2(t), …, S n (t). The fused time series S fusion (t) is calculated by the formula where ω i (t) is the weight of the i-th time series at time t. The determination method of the weight ω i (t) is as follows: First, calculate the correlation coefficient r i j(t) between each time series and other time series. The formula is where and are the means of the corresponding time series within the time window [t - τ, t]. Then calculate the weight according to the correlation coefficient This algorithm can dynamically adjust the weight according to the correlation of time series data at different times, effectively fuse multi-view time series data, and improve the accuracy and adaptability of time series data fusion.

[0017] Furthermore, data interaction between the sub-devices is carried out through a hybrid communication method that combines quantum communication and traditional communication. Under normal circumstances, quantum communication is preferentially used for rapid synchronization of multi-perspective data, and the long-distance effect of quantum entanglement states is utilized to achieve near-real-time data transmission. When quantum communication fails or is interfered with, it automatically switches to the traditional communication method. The switching mechanism is triggered by real-time monitoring of the bit error rate (BER) and signal strength (RSSI) of the quantum communication link. When BER > BER th and RSSI < RSSI th At this time, BER th and RSSI th are preset thresholds determined according to experimental tests. The switching process is initiated. During the switching process, data continuity is ensured through a caching mechanism. The untransmitted quantum communication data is temporarily stored in the local cache and continued to be transmitted after switching to the traditional communication method. This hybrid communication method improves the reliability and stability of device data interaction, ensuring the continuous progress of multi-perspective data fusion work in complex environments.

[0018] Furthermore, the device also has a self-learning and optimization function. During the multi-perspective data fusion process, by continuously comparing the fusion results with the actual scenario data, the parameters in the fusion algorithm are adjusted and optimized using an optimization algorithm based on reinforcement learning. Let the parameters in the fusion algorithm be θ, and the optimization goal is to minimize the error E between the fusion result and the actual scenario data. The Q-learning algorithm is adopted. The state space S is the parameter combination of the fusion algorithm, and the action space A is the adjustment operation for the parameters. At each time step t, according to the current state s t Select action a t , and after executing the action, a new state s t+1 and reward r t are obtained. The reward r t =-λ×E(s t+1 ), where λ is the reward coefficient determined according to experiments. By continuously iteratively updating the Q-value table, the formula is where α is the learning rate and γ is the discount factor, both determined by experimental optimization. Finally, the optimal parameter adjustment strategy is selected according to the optimized Q-value table, continuously improving the accuracy and effect of multi-perspective data fusion.

[0019] Compared with the prior art, the multi-perspective modeling data fusion device has the following beneficial effects:

[0020] 1. The device combines quantum entanglement, quantum communication with advanced fusion algorithms to achieve efficient and accurate fusion of multi-perspective data. Utilizing the long-distance effect of quantum entanglement, the device can achieve near-real-time data transmission and synchronization, significantly improving the timeliness of data fusion. At the same time, by introducing a quantum error correction mechanism and advanced fusion algorithms, the device can effectively reduce the bit error rate during data transmission, ensuring the accuracy and reliability of the fusion results, which provides strong technical support for constructing a more comprehensive and accurate multi-perspective model.

[0021] 2. The device also has self-learning and optimization functions and can continuously improve itself in practical applications. By comparing the fusion results with the actual scenario data, the device uses reinforcement learning algorithms to adjust and optimize the parameters in the fusion algorithms, thereby improving the accuracy and effect of data fusion. This adaptive learning and optimization mechanism enables the device to make flexible adjustments according to different application scenarios and data characteristics, better meeting the actual needs. In addition, the device adopts a hybrid communication method to ensure the reliability and stability of data interaction, and can ensure the continuous progress of multi-perspective data fusion work even in complex environments.

[0022] Other advantages, objectives and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0024] Figure 1 It is a functional implementation flowchart of a multi-perspective modeling data fusion device;

[0025] Figure 2 It is an overall architecture flowchart of a multi-perspective modeling data fusion device. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention objectives, the following will, in conjunction with the accompanying drawings and preferred embodiments, describe in detail the specific embodiments, structures, features and their effects of the present invention as follows.

[0027] Embodiment 1

[0028] This embodiment describes that in an intelligent transportation system, it is necessary to fuse multi-perspective data from road cameras (image data) and vehicle sensors (time series data) to achieve accurate traffic flow monitoring and intelligent scheduling.

[0029] The data acquisition module of the sub-device is connected to the road camera and the vehicle sensor. For the speed and acceleration time series data collected by the vehicle sensor, an adaptive sampling algorithm is used for acquisition. Suppose the data of a vehicle speed sensor is the time series x(t), the initial sampling frequency f s0 = 10Hz, the frequency adjustment coefficient α = 0.5, the number of data points n in the sampling window = 20. At a certain moment, the mean value of the data in the window The standard deviation σ = 3m / s. According to the formula Calculate the sampling frequency f at this time s , so as to dynamically adjust the sampling frequency and obtain data more efficiently.

[0030] For the image data collected by the road camera, in the data encoding stage, no processing is performed first (because subsequent mainly focuses on image fusion algorithms). For the numerical data of the vehicle sensor, such as the speed value d = 30m / s, assume that the speed data range is divided into m = 5 intervals, and each interval corresponds to a quantum bit superposition state. The central value c i and the width correlation coefficient σ i Are set according to the actual speed range. According to the formula Determine the corresponding quantum bit superposition state and load the data information onto the entangled state quantum bits.

[0031] Each sub-device (such as monitoring devices deployed in different sections) shares data through the quantum entanglement module. When a sub-device needs to share data with other sub-devices, by operating on the local entangled state quantum bits, using the non-local effect of quantum entanglement, the quantum bits of the remote sub-device undergo corresponding state changes. After receiving the quantum bits with state changes, the original multi-perspective data is restored through decoding operations. At the same time, a quantum error correction mechanism is introduced. If an error is detected, assume the original entangled state quantum bit pair sent by the sender is |ψ AB >>, and an additional auxiliary entangled state quantum bit pair |ψ CD >> is sent. The receiver calculates the error amount according to the formula Δ = |<ψ' AB |ψ AB >-<ψ' CD |ψ CD >| and corrects it according to the pre-established error correction table.

[0032] Taking the images I1, I2,..., I n Collected by multiple road cameras as input, perform multi-scale decomposition on each image to obtain feature maps F at different scalesij , calculate the attention weight w of each feature map ij , and the formula is where g is the feature extraction function, is the k-th feature channel of the feature map F i j, θ k is the feature channel weight, and the finally fused image feature In this way, important regions can be highlighted to improve the understanding of traffic scenes.

[0033] The speed time series data of multiple vehicles S1(t), S2(t), …, S n (t), the fused time series First, calculate the correlation coefficient between each time series and other time series where and are the means of the corresponding time series within the time window [t - τ, t], and then calculate the weight according to the correlation coefficient Realize the dynamic fusion of time series data to provide more accurate data support for traffic flow prediction.

[0034] Data interaction between each sub-device is carried out through a hybrid communication method that combines quantum communication and traditional communication. Under normal circumstances, quantum communication is preferentially used for fast synchronization. When the bit error rate BER of the quantum communication link > BER th and the signal strength RSSI < RSSI th , it automatically switches to the traditional communication method and ensures the continuity of data through a caching mechanism.

[0035] During the multi-view data fusion process, continuously compare the fusion result (such as the traffic flow prediction value) with the actual traffic flow data. Use an optimization algorithm based on reinforcement learning to adjust and optimize the parameters in the fusion algorithm. Let the parameters in the fusion algorithm be θ, and adopt the Q-learning algorithm. The state space S is the parameter combination of the fusion algorithm, and the action space A is the adjustment operation of the parameters. At each time step t, according to the current state s t select the action a t , and after executing the action, obtain the new state s t +1 and the reward r t = -λ × E(s t+1 ), where λ is the reward coefficient. By continuously iterating and updating the Q-value table Q(s t , a t ) = Q(s t , a t ) + α × (r t + γ × maxQ(s t+1 , at+1 ) - Q(s t , a t ))), where α is the learning rate and γ is the discount factor. Finally, optimize the fusion algorithm to improve the accuracy of traffic flow monitoring and prediction.

[0036] Embodiment 2

[0037] This embodiment describes that on an industrial production line, it is necessary to fuse multi - perspective data of equipment sensors (time - series data) and production - environment cameras (image data) to achieve real - time monitoring of the production process and fault warning.

[0038] The data acquisition module of the sub - device is connected to the equipment sensor and the production - environment camera. For the temperature and pressure time - series data collected by the equipment sensor, use the adaptive sampling algorithm. For example, for the time - series x(t) of a certain temperature sensor data, the initial sampling frequency f s0 = 5Hz, the frequency adjustment coefficient α = 0.3, and the number of data points n in the sampling window = 15. At a certain moment, the mean value of the data in the window The standard deviation σ = 2°C. According to the formula Adjust the sampling frequency to ensure that key data changes are captured.

[0039] For the image data collected by the production - environment camera, no encoding processing is done first (the subsequent focus is on image fusion). For the numerical data of the equipment sensor, such as the pressure value d = 5MPa, if the pressure data range is divided into m = 4 intervals, determine the central value c i and the width - related coefficient σ i , and encode according to the formula and load the data onto the entangled quantum bits.

[0040] Each sub - device (distributed at different positions on the production line) shares data through the quantum entanglement module. Utilize the non - local effect of quantum entanglement to achieve rapid data synchronization. If a quantum bit state error is detected, adopt a quantum error - correction mechanism. Through the formula Δ = |<ψ' AR |ψ AB > - <ψ' CD |ψ CD >| calculate the error amount and correct it.

[0041] Take the images I1, I2, …, I n collected by multiple production - environment cameras as input, perform multi - scale decomposition on the images to obtain the feature map F ij , calculate the attention weights The finally fused image features Help the operator observe the equipment operation status and product quality on the production line more clearly.

[0042] The time series data S1(t), S2(t), …, S n (t) of multiple device sensors, and the fused time series First, calculate the correlation coefficient Then, calculate the weights Realize the dynamic fusion of time series data and promptly detect abnormal device operations.

[0043] The sub-devices adopt a combination of quantum communication and traditional communication. When normal, quantum communication is used. When the bit error rate BER of the quantum communication link > BER th and the signal strength RSSI < RSSI th , switch to traditional communication and use a caching mechanism to ensure data continuity.

[0044] During the multi-view data fusion process, compare the fusion results (such as device fault prediction results) with the actual production situation, and use an optimization algorithm based on reinforcement learning to adjust the fusion algorithm parameters. Adopt the Q-learning algorithm, where the state space S is the combination of fusion algorithm parameters, the action space A is the parameter adjustment operation. At each time step t, according to the state s t Select the action a t , obtain the new state s t +1 and the reward r t = -λ × E(s t+1 ), and optimize the fusion algorithm by iteratively updating the Q-value table to improve the accuracy of fault warning.

[0045] The above is only a preferred embodiment of the present invention and does not impose any form of limitation on the present invention. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or decorations equivalent to changes within the scope of the technical solution of the present invention without departing from the technical solution of the present invention. However, any brief modifications, equivalent changes, and decorations made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A multi-view modeling data fusion device, characterized in that The device comprises the following components: a quantum entanglement module, data encoding and modulation, quantum communication and synchronization, and fusion algorithms and processing; The quantum entanglement module: Inside each sub-device, a quantum entanglement generation and control module is integrated. This module can generate pairs of entangled qubits and allocate a part of the qubits to other sub-devices that need to work collaboratively; The data encoding and modulation: At the sending end, the sub-device encodes the multi-view data collected, converts it into a form of information that can be carried by qubits, and uses quantum modulation technology to load the data information onto the entangled qubits, enabling them to carry the characteristics of multi-view data; The quantum communication and synchronization: When a sub-device needs to share multi-view data with other sub-devices, by operating on the local entangled qubits, according to the non-local effect of quantum entanglement, the qubits of the remote sub-device entangled with them will instantaneously undergo corresponding state changes. After the remote sub-device receives the qubits with state changes, the original multi-view data is restored through decoding operations; The fusion algorithms and processing: After each sub-device receives the multi-view data synchronized from other sub-devices, it uses fusion algorithms to perform fusion processing on the local data and the remotely synchronized data.

2. The multi-perspective modeling data fusion device according to claim 1, wherein, The data acquisition module of the sub-device includes various types of sensor interfaces. For different types of sensor data, the sub-device uses an adaptive sampling algorithm for data acquisition. The algorithm is as follows: Let the sensor data be the time series x(t), and the sampling frequency f s is adjusted according to the formula where f s0 is the initial sampling frequency, α is the frequency adjustment coefficient, n is the number of data points within the sampling window, is the mean value of the data within the window, and σ is the standard deviation of the data within the window.

3. The multi-view modeling data fusion device according to claim 1, characterized in that The quantum entanglement generation and control module generates entangled state qubit pairs through an improved spontaneous parametric down-conversion technology based on nonlinear optical processes. Specifically, an additional phase modulator is introduced, and its modulation parameter φ is determined by the formula where P in is the input laser power, P th is the threshold power, and β is the modulation coefficient.

4. A multi-perspective modeling data fusion device according to claim 1, wherein In the data encoding stage, a dynamic encoding algorithm based on the superposition state of qubits is adopted. For the numerical data in the multi-view data, that is, the sensor measurement values, the data range is divided into m intervals, and each interval corresponds to a superposition state of qubits. Let the data value be d, and its corresponding qubit superposition state |ψ> is determined by the formula where a i is the superposition coefficient, c i is the central value of the i-th interval, and σ i is the width correlation coefficient of the i-th interval. For the text data, it is first converted into word vectors through natural language processing, and then the numerical values of each dimension of the word vectors are encoded according to the above numerical data encoding method.

5. The multi-view modeling data fusion device according to claim 1, wherein In the quantum communication and synchronization process, a quantum error correction mechanism is introduced. When an error in the quantum bit state is detected, an error correction algorithm based on quantum entanglement redundancy is adopted. Let the original entangled state quantum bit pair sent by the sending end be |ψ AB >, and at the same time, an additional pair of auxiliary entangled state quantum bit pairs |ψ CD > is sent. After the receiving end receives the quantum bits, the quantum bit state information is obtained through a measurement operation. If an error is detected, according to the formula Δ = |<ψ' AB |ψ AB > - <ψ' CD |ψ CD >|, the error amount is calculated, where |ψ AB ’| and |ψ CD ’| are the quantum bit states actually received by the receiving end. Then, according to the pre-established error correction table, a correction operation is performed on the erroneous quantum bit state.

6. The multi-view modeling data fusion device according to claim 1, characterized in that, In the data fusion stage, for the fusion of image data in multi-view data, an algorithm based on multi-scale feature fusion and attention mechanism is adopted. Let the input multi-view images be I1, I2, …, I n , first perform multi-scale decomposition on each image to obtain feature maps F ij at different scales, where i represents the image number and j represents the scale level. Then calculate the attention weight w ij of each feature map. The formula is w ij = where g is the feature extraction function, is the k-th feature channel of the feature map F ij , θ k is the feature channel weight. Finally, the fused image feature F fusion is 7. A multi-view modeling data fusion device according to claim 1, characterized in that, In the data fusion stage, for the fusion of time series data in multi-view data, a dynamic fusion algorithm based on adaptive weights is adopted. Let multiple time series data be S1(t), S2(t), …, S n (t), and the fused time series S fusion (t) is calculated by the formula where ω i (t) is the weight of the i-th time series at time t, and the determination method of the weight ω i (t) is as follows: First, calculate the correlation coefficient r i j(t) between each time series and other time series, and the formula is where and are the means of the corresponding time series within the time window [t - τ, t]. Then, calculate the weight according to the correlation coefficient This algorithm can dynamically adjust the weights according to the correlation of time series data at different times.

8. A multi-view modeling data fusion device according to claim 1, characterized in that Data interaction between the sub-devices is carried out through a hybrid communication method that combines quantum communication and traditional communication. Under normal circumstances, quantum communication is preferentially used for rapid synchronization of multi-perspective data, and the long-distance effect of quantum entanglement states is utilized to achieve near-real-time data transmission. When quantum communication fails or is interfered with, it automatically switches to the traditional communication method. The switching mechanism is triggered by real-time monitoring of the bit error rate (BER) and received signal strength indication (RSSI) of the quantum communication link. When BER > BER th and RSSI < RSSI th the switching process is initiated. During the switching process, data continuity is ensured through a caching mechanism. Quantum communication data that has not been completely transmitted is temporarily stored in the local cache and continues to be transmitted after switching to the traditional communication method.

9. The multi-view modeling data fusion device according to claim 1, wherein The device also has a self-learning and optimization function. During the multi-view data fusion process, by continuously comparing the fusion result with the actual scene data, the parameters in the fusion algorithm are adjusted and optimized using an optimization algorithm based on reinforcement learning. Let the parameters in the fusion algorithm be θ, and the optimization goal is to minimize the error E between the fusion result and the actual scene data. The Q-learning algorithm is adopted. The state space S is the parameter combination of the fusion algorithm, and the action space A is the adjustment operation of the parameters. At each time step t, according to the current state s t select an action a t , and after executing the action, a new state s t+1 and a reward r t are obtained. The reward r t = -λ × E(s t+1 ), where λ is the reward coefficient. By continuously iteratively updating the Q-value table, the formula is where α is the learning rate and γ is the discount factor. Finally, the optimal parameter adjustment strategy is selected according to the optimized Q-value table.

Images