Data processing method and system based on quantum deep learning

Abstract

The invention discloses a data processing method and system based on quantum deep learning, and the method comprises the steps: collecting sensor data, equipment logs and environment parameters of production equipment in real time, and constructing a multi-dimensional time series data set through the collected data; mapping the multi-source data into a superposition state of quantum bits through a tunable quantum encoder so as to fuse operation data of different devices at the same time; and constructing a quantum security detection model. During implementation, industrial-grade quantum coding: a tunable quantum encoder supports unified quantum state coding of multi-source heterogeneous data, such as scalar, time sequence and image, and the coding efficiency is also remarkably improved; according to the quantum anomaly scoring mechanism, a quantum anomaly scoring layer quantifies the equipment state deviation degree based on quantum state entanglement and measurement results, and the detection precision is remarkably improved. And security threshold optimization: predicting the performance attenuation trend of the equipment in combination with the quantum Bayesian network, and dynamically correcting the security threshold range.

Description

Technical Field

[0001] The present application relates to the field of industrial safety monitoring technology, and specifically to a data processing method and system based on quantum deep learning. Background Art

[0002] In industrial production, the safe operation of equipment is the core of ensuring production safety. Traditional methods rely on sensor data threshold alarms or anomaly detection based on classic machine learning (such as SVM, random forest), but there are the following problems:

[0003] High data dimension: Time series data (such as temperature, vibration, and pressure) from multiple devices and sensors has high dimensions, and the processing efficiency of classic models is low;

[0004] Large noise interference: Electromagnetic interference and sensor noise in industrial environments can easily lead to false alarms;

[0005] Poor dynamic adaptability: The operating status of equipment changes dynamically, and traditional models are difficult to adaptively adjust safety thresholds.

[0006] However, the use of quantum computing and deep learning to ensure the safe operation of equipment is still in a blank stage. For this reason, this application proposes a data processing method and system based on quantum deep learning. Summary of the invention

[0007] To this end, the present application provides a data processing method and system based on quantum deep learning to solve the problems of low processing efficiency, large noise interference, and poor dynamic adaptability of the classical model in the prior art.

[0008] In order to achieve the above objectives, this application provides the following technical solutions:

[0009] A data processing method based on quantum deep learning, comprising: collecting sensor data, equipment logs and environmental parameters of production equipment in real time, and constructing a multi-dimensional time series data set through the collected data;

[0010] Map multi-source data into the superposition state of quantum bits through a tunable quantum encoder to merge the operating data of different devices at the same time;

[0011] Construct a quantum safety detection model, which includes a quantum convolution layer, a quantum long short-term memory unit, and a quantum anomaly scoring layer, wherein the quantum anomaly scoring layer determines whether the state of the computing device deviates from the safety threshold range;

[0012] Through hybrid training and dynamic optimization, a classical-quantum hybrid optimizer is used to train the model based on the historical security data of the device to dynamically adjust the quantum gate parameters and security thresholds;

[0013] Real-time safety judgment and early warning, triggering graded alarm signals based on the abnormal scores output by the model, inputting real-time data into the model, and triggering graded alarm signals if the abnormal scores exceed the safety threshold range. Graded alarm signals include early warning and emergency shutdown.

[0014] Preferably, the tunable quantum encoder realizes data encoding through parameterized rotating gates and controlled entanglement gates, and the number of encoding bits is dynamically allocated according to the data dimension.

[0015] Preferably, the quantum anomaly scoring layer generates a device anomaly probability score by calculating the trace distance between the quantum state density matrix and a preset safety reference state.

[0016] Preferably, the safety threshold adjustment is based on a quantum Bayesian network, the input includes historical equipment performance data, environmental parameters and maintenance records, and the output is an adaptive safety threshold range.

[0017] Preferably, an error mitigation unit is embedded in the quantum security detection model, and the error mitigation unit suppresses quantum noise through surface code quantum error correction technology and optimizes quantum gate parameters using a dynamic annealing algorithm.

[0018] A data processing system based on quantum deep learning, comprising:

[0019] Data acquisition module, used for real-time acquisition of multi-protocol industrial data;

[0020] Quantum coding module, which performs quantum state mapping of multi-source data;

[0021] Quantum computing core, running quantum safety detection models;

[0022] Security threshold management module, to achieve adaptive adjustment of security threshold;

[0023] The alarm execution module links the industrial control system to perform hierarchical response.

[0024] Preferably, the quantum computing core supports collaborative computing between a quantum processor and a classical GPU, wherein the quantum convolution layer and the quantum long short-term memory unit run on the quantum processor, and the quantum anomaly scoring layer accelerates computing on the GPU.

[0025] Preferably, the alarm execution module includes a three-level response mechanism:

[0026] Level 1 response: When the anomaly score exceeds the threshold by 10%, an early warning signal is sent to the monitoring terminal;

[0027] Secondary response: When the abnormality score exceeds the threshold by 30%, the equipment is triggered to reduce load;

[0028] Level 3 response: When the abnormality score exceeds the threshold by 50%, an emergency shutdown command is executed.

[0029] Preferably, the data acquisition module supports Modbus, Profinet and MQTT protocols, and the data acquisition module integrates a data cleaning unit for eliminating abnormal sampling points.

[0030] Preferably, the system provides OPC UA and REST API dual interfaces, which are compatible with the industrial Internet of Things platform and the cloud-edge collaborative architecture.

[0031] Compared with the prior art, this application has at least the following beneficial effects:

[0032] Industrial-grade quantum coding: Tunable quantum encoders support unified quantum state coding of multi-source heterogeneous data, such as scalars, time series, and images, and the coding efficiency has also been significantly improved;

[0033] Quantum anomaly scoring mechanism: The quantum anomaly scoring layer quantifies the degree of device state deviation based on quantum state entanglement and measurement results, and the detection accuracy is significantly improved;

[0034] Safety threshold optimization: Combined with quantum Bayesian network to predict the performance degradation trend of equipment and dynamically correct the safety threshold range;

[0035] Noise-resistant design: Error mitigation units are embedded in quantum circuits to suppress industrial environmental noise through quantum error correction codes. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more intuitively illustrate the prior art and the present application, exemplary drawings are given below. It should be understood that the specific shapes and structures shown in the drawings should not generally be regarded as limiting conditions for implementing the present application; for example, those skilled in the art are capable of easily making conventional adjustments or further optimizations to the addition / reduction / attribution division, specific shapes, positional relationships, connection methods, dimensional ratios, etc. of certain units (components) based on the technical concepts and exemplary drawings disclosed in the present application.

[0037] Figure 1 A flowchart of a data processing method based on quantum deep learning provided in Example 1 of the present application;

[0038] Figure 2 This is a module diagram of the data processing system based on quantum deep learning provided in Example 1 of the present application. DETAILED DESCRIPTION

[0039] The present application is further described below in detail through specific embodiments in conjunction with the accompanying drawings.

[0040] like Figure 1As shown, a data processing method based on quantum deep learning includes: real-time collection of sensor data, equipment logs and environmental parameters of production equipment; the operating status of the equipment can be obtained through the sensor data, and the past operating data of the equipment can be obtained through the equipment log. When the equipment is running, the environmental parameters in which it is located will also have a certain impact on the equipment. By obtaining the above data, the situation of the equipment can be fully grasped, so as to judge the status of the equipment;

[0041] Mapping multi-source data into superposition states of quantum bits through a tunable quantum encoder;

[0042] Construct a quantum security detection model, which includes a quantum convolution layer, a quantum long short-term memory unit, and a quantum anomaly scoring layer, so as to accurately and quickly process and evaluate data;

[0043] The classical-quantum hybrid optimizer training model is used to dynamically adjust quantum gate parameters and safety thresholds. That is, during implementation, the safety threshold is adjusted according to the parameters of the on-site working conditions and the coordination between different devices. By adjusting the safety threshold of the device, the fixed threshold in the traditional technical solution is replaced. Therefore, the safety threshold of the device can be adjusted according to different working conditions and the coordination between different devices, ensuring that the operation of the device can be safer.

[0044] The abnormal scores output by the model trigger graded alarm signals, thereby giving different safety warning signals to guide users to take timely actions. By generating alarm signals of different levels, users can make a rough judgment on the warning situation of the equipment, so that they can better handle the warning.

[0045] Specifically include the following steps:

[0046] Step 1: Multi-source equipment data collection, real-time acquisition of sensor data, equipment logs and environmental parameters of production equipment, and construction of multi-dimensional time series data sets, so as to more accurately judge the status of the equipment;

[0047] Step 2: Quantum state encoding and feature fusion, mapping multi-source data into the superposition state of quantum bits through a tunable quantum encoder, integrating the operating data of different devices at the same time;

[0048] Step 3: Build a quantum safety detection model, which includes a quantum convolution layer, a quantum long short-term memory unit, and a quantum anomaly scoring layer. The quantum anomaly scoring layer determines whether the computing device state deviates from the safety threshold range. The quantum long short-term memory unit combines the characteristics of quantum computing to improve the model's ability to process time series data, thereby more effectively determining whether the device is in a safe operating state.

[0049] Step 4: Hybrid training and dynamic optimization, using a classical-quantum hybrid optimizer to train the model based on the historical security data of the device, ensuring that the trained model can better match the actual state of the device, thereby dynamically adjusting the quantum gate parameters and security threshold range;

[0050] Step 5: Real-time safety judgment and warning. Input real-time data into the model. If the abnormal score exceeds the safety threshold range, a graded alarm signal is triggered. The graded alarm signal includes warning and emergency shutdown (of course, other different levels of alarm signals can also be set according to user needs).

[0051] The tunable quantum encoder realizes data encoding through parameterized rotating gates and controlled entanglement gates. The number of encoding bits is dynamically allocated according to the data dimension. Data encoding through parameterized rotating gates and controlled entanglement gates can be encoded more concisely and efficiently, so as to ensure that the scheme can be more stable and efficient when implemented.

[0052] The quantum anomaly scoring layer generates a device anomaly probability score by calculating the trace distance between the quantum state density matrix and the preset safety reference state, thereby making the score more accurate and reducing the occurrence of scoring errors.

[0053] The safety threshold adjustment is based on a quantum Bayesian network, and the input includes historical performance data of the equipment, environmental parameters and maintenance records, which ensures the authenticity of the data, and the output is an adaptive safety threshold range.

[0054] An error mitigation unit is embedded in the quantum security detection model. The error mitigation unit suppresses quantum noise through surface code quantum error correction technology and uses a dynamic annealing algorithm to optimize quantum gate parameters, further reducing the occurrence of false alarms and ensuring the safe and stable operation of the equipment.

[0055] like Figure 2 As shown, a data processing system based on quantum deep learning includes:

[0056] Data acquisition module: Integrates industrial sensors, equipment PLC interface and environmental monitoring unit, supports real-time data acquisition of multiple protocols, can adapt to a variety of interfaces, and is used to collect data from different devices, making the system more widely used;

[0057] Quantum coding module: Achieve efficient mapping of data to quantum states through tunable quantum encoders, supporting dynamic bit allocation;

[0058] Quantum computing core: A quantum processor or quantum simulator runs a quantum safety detection model, which includes a reconfigurable quantum circuit. After collecting data, the safety detection model can perform safety detection on the data to determine whether the device can continue to operate safely.

[0059] Safety threshold management module: The safety threshold is adaptively adjusted according to the equipment aging coefficient and environmental changes. Different equipment and equipment with different usage time have different safety thresholds. Through the safety threshold management module, different safety thresholds can be set according to different equipment and the situations between different equipment, thus ensuring the safe operation of the equipment.

[0060] Alarm execution module: Links with industrial control systems (such as DCS, SCADA) to execute hierarchical alarms or shutdown instructions.

[0061] The quantum computing core supports the collaborative computing of quantum processors and classical GPUs, wherein the quantum convolution layer and quantum long short-term memory unit run on the quantum processor, and the quantum anomaly scoring layer accelerates computing on the GPU.

[0062] As one of the implementation methods of the alarm execution module:

[0063] The alarm execution module includes a three-level response mechanism:

[0064] Level 1 response: When the anomaly score exceeds the threshold by 10%, an early warning signal is sent to the monitoring terminal;

[0065] Secondary response: When the abnormality score exceeds the threshold by 30%, the equipment is triggered to reduce load;

[0066] Level 3 response: When the abnormality score exceeds the threshold by 50%, an emergency shutdown command is executed.

[0067] The data acquisition module supports Modbus, Profinet and MQTT protocols. The data acquisition module integrates a data cleaning unit for eliminating abnormal sampling points to avoid the impact of abnormal sampling points on the safe operation of the equipment.

[0068] The system provides dual interfaces of OPC UA and REST API for compatibility with industrial Internet of Things platforms and cloud-edge collaborative architecture.

[0069] The technical features of the above embodiments may be arbitrarily combined (as long as there is no contradiction in the combination of these technical features). To make the description concise, not all possible combinations of the technical features in the above embodiments are described; these embodiments that are not explicitly written should also be considered to be within the scope of this specification.

Claims

1. A data processing method based on quantum deep learning, characterized in that: include: Collect sensor data, equipment logs, and environmental parameters of production equipment in real time, and build a multi-dimensional time series data set based on the collected data; Map multi-source data into the superposition state of quantum bits through a tunable quantum encoder to merge the operating data of different devices at the same time; Construct a quantum safety detection model, which includes a quantum convolution layer, a quantum long short-term memory unit, and a quantum anomaly scoring layer, wherein the quantum anomaly scoring layer determines whether the state of the computing device deviates from the safety threshold range; Through hybrid training and dynamic optimization, a classical-quantum hybrid optimizer is used to train the model based on the historical security data of the device to dynamically adjust the quantum gate parameters and security thresholds; Real-time safety judgment and early warning, triggering graded alarm signals based on the abnormal scores output by the model, inputting real-time data into the model, and triggering graded alarm signals if the abnormal scores exceed the safety threshold range. Graded alarm signals include early warning and emergency shutdown.

2. According to claim 1, a data processing method based on quantum deep learning is characterized in that: The tunable quantum encoder realizes data encoding through parameterized rotation gates and controlled entanglement gates, and the number of encoding bits is dynamically allocated according to the data dimension.

3. The data processing method based on quantum deep learning according to claim 1, characterized in that: The quantum anomaly scoring layer generates a device anomaly probability score by calculating the trace distance between the quantum state density matrix and the preset safety reference state.

4. The data processing method based on quantum deep learning according to claim 1, characterized in that: The adjustment of the safety threshold is based on a quantum Bayesian network, the input includes historical performance data of the equipment, environmental parameters and maintenance records, and the output is an adaptive safety threshold range.

5. The data processing method based on quantum deep learning according to claim 1, characterized in that: An error mitigation unit is embedded in the quantum security detection model. The error mitigation unit suppresses quantum noise through surface code quantum error correction technology and uses a dynamic annealing algorithm to optimize quantum gate parameters.

6. A data processing system based on quantum deep learning, characterized in that: include: Data acquisition module, used for real-time acquisition of multi-protocol industrial data; Quantum coding module, which performs quantum state mapping of multi-source data; Quantum computing core, running quantum safety detection models; Security threshold management module, to achieve adaptive adjustment of security threshold; The alarm execution module links the industrial control system to perform hierarchical response.

7. A data processing system based on quantum deep learning according to claim 6, characterized in that: The quantum computing core supports the collaborative computing of quantum processors and classical GPUs, wherein the quantum convolution layer and quantum long short-term memory unit run on the quantum processor, and the quantum anomaly scoring layer accelerates computing on the GPU.

8. The data processing system based on quantum deep learning according to claim 6, characterized in that: The alarm execution module includes a three-level response mechanism: Level 1 response: When the anomaly score exceeds the threshold by 10%, an early warning signal is sent to the monitoring terminal; Secondary response: When the abnormality score exceeds the threshold by 30%, the equipment is triggered to reduce load; Level 3 response: When the abnormality score exceeds the threshold by 50%, an emergency shutdown command is executed.

9. The data processing system based on quantum deep learning according to claim 6, characterized in that: The data acquisition module supports Modbus, Profinet and MQTT protocols, and the data acquisition module integrates a data cleaning unit for eliminating abnormal sampling points.

10. A data processing system based on quantum deep learning according to claim 6, characterized in that: The system provides dual interfaces of OPC UA and REST API for compatibility with industrial Internet of Things platforms and cloud-edge collaborative architecture.

Images