Artificial intelligence-based historic site monitoring method and system

Abstract

The invention relates to the technical field of signal processing, and provides an ancient site monitoring method and system based on artificial intelligence, and the method comprises the steps: obtaining the multi-dimensional feature data of a to-be-monitored ancient site; generating an intelligent monitoring scheme through a preset scheme planning model according to the multi-dimensional feature data; monitoring equipment and an alarm are deployed according to the intelligent monitoring scheme, and working parameters are set; and executing a monitoring process in the intelligent monitoring scheme, controlling each monitoring device to cooperatively monitor the historic site, and starting an alarm to give an alarm according to an alarm strategy after abnormal data is monitored. According to the method, the actual condition of the to-be-monitored historic site is comprehensively reflected through the multi-source data, the scheme planning model can efficiently and accurately determine the degenerated disease condition of the historic site, then an intelligent monitoring scheme is planned in a targeted mode, closed-loop management from data collection to alarm response is achieved, efficient and accurate monitoring can be conducted on the historic site, and the monitoring efficiency is improved. And the method has good universality and can meet monitoring requirements of historic sites in various regions.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular to an artificial intelligence-based ancient monument monitoring method and an artificial intelligence-based ancient monument monitoring system. Background Art

[0002] Murals and stone carvings are an important part of human cultural heritage and have extremely high historical, artistic and scientific value. However, these monuments are facing multiple threats such as natural erosion and human destruction. Traditional monitoring methods have problems such as low efficiency and poor accuracy, which makes it difficult to meet the comprehensive and real-time protection needs of murals and stone carvings.

[0003] Related technologies generally rely on visible light cameras, and only use simple image acquisition to judge the development status of murals and stone carvings based on image color changes. Single parameter monitoring cannot reflect the multi-factor coupling effect of cultural relics degradation, it is difficult to detect internal cracks and microbial activities, and it is impossible to quantify the correlation between environmental factors and damage evolution, so it is impossible to carry out comprehensive and real-time protection according to the actual situation of murals and stone carvings. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a historical monument monitoring method based on artificial intelligence.

[0005] The invention also proposes a historical site monitoring system based on artificial intelligence.

[0006] To achieve the above-mentioned purpose, the first aspect of the present invention proposes an artificial intelligence-based ancient monument monitoring method, comprising the following steps: S1, obtaining multi-dimensional feature data of the ancient monument to be monitored, the multi-dimensional feature data including mural surface image data, ancient monument space environment parameters and supporting structure deformation data; S2, generating an intelligent monitoring scheme through a preset scheme planning model according to the multi-dimensional feature data, the intelligent monitoring scheme including: at least one type of monitoring equipment and its layout position, layout method and working parameters; at least one alarm and its layout position, layout method, alarm strategy and working parameters; monitoring process of the monitoring equipment; the scheme planning model is a machine learning model obtained by iterative training of historical monitoring scheme matching data; S3, deploying monitoring equipment and alarms according to the intelligent monitoring scheme, and setting working parameters; S4, executing the monitoring process in the intelligent monitoring scheme, controlling each monitoring device to coordinately monitor the ancient monument, and enabling the alarm to alarm according to the alarm strategy after abnormal data is monitored.

[0007] The above-mentioned artificial intelligence-based monument monitoring method of the present invention may also have the following additional technical features: According to one embodiment of the present invention, the monitoring equipment includes: a three-dimensional laser scanning module, a multi-spectral imaging module, a macro camera, a video acquisition camera, a non-contact vibration sensor, a millimeter wave radar, a fiber Bragg grating sensor, an environmental sensor, an infrared thermal imager, a particle counter, an ultraviolet irradiance meter, an acoustic emission sensor, a gas composition analyzer, an air microbial sampler, a biosensor and a structural health monitoring system.

[0008] According to one embodiment of the present invention, the alarm includes: a directional sound wave alarm, a visual warning device and a rapid response protective cover at the physical end; and a pre-set multi-level notification system and an alarm log system at the data end.

[0009] According to one embodiment of the present invention, the step S1 of obtaining multi-dimensional feature data of the monitored monument specifically includes the following steps: S11, obtaining three-dimensional point cloud data of the monument surface with millimeter-level accuracy through three-dimensional laser scanning; S12, using a multi-spectral imaging device to collect spectral reflectance data of the mural surface, and merging it with the three-dimensional point cloud data to generate mural surface image data; S13, deploying a micro-environment sensor array, continuously collecting environmental parameters, and generating monument space environmental parameters; environmental parameters include temperature, humidity, light intensity, air flow rate and VOC (Volatile Organic Compounds) concentration parameters; S14, retrieving historical monitoring data and restoration records of the monument through a historical data interface, and extracting support structure deformation data.

[0010] According to one embodiment of the present invention, the above-mentioned artificial intelligence-based ancient monument monitoring method also includes: S5, acquiring monitoring data in real time and when abnormal data appears in the monitoring data that meets the preset dynamic optimization conditions, the abnormal data and the monitoring data in the previous optimization cycle are packaged and input into the scheme planning model to update the optimized intelligent monitoring scheme.

[0011] According to one embodiment of the present invention, the training method of the scheme planning model includes the following steps: constructing a historical scheme library with weight annotations, the annotation dimensions include the disease detection efficiency weight, the equipment energy consumption weight and the data completeness weight, the three weights are set by the management personnel and the sum of the weights is 100%; using the Wasserstein GAN (Generative Adversarial Networks) architecture for adversarial training, the generator input is multi-dimensional feature data, and the discriminator evaluates the distribution consistency of the output scheme and the expert scheme; adapting the pre-trained model to the new monument scene through transfer learning, and the number of fine-tuning layers does not exceed 20% of the total number of layers.

[0012] According to one embodiment of the present invention, the monitoring process includes a corrosion risk warning sub-process, and the corrosion risk warning sub-process specifically includes the following steps: calculating the corrosion index CI according to the gas composition in the monitored environmental parameters, CI=0.35×[SO 2 ]+0.28×[NO 2 ]+0.37×[Cl - ], where CI is the corrosion index, [SO 2 ] is the concentration of sulfur dioxide in the air, [NO 2 ] is the concentration of nitrogen dioxide in the air, [Cl - ] is the chloride ion concentration in the air; the humidity data RH in the environmental parameters is obtained. When CI>0.85 and RH>65%, it is determined whether there is a quick response protective cover in the area. If so, the quick response protective cover is activated, the gas sampling frequency is increased to the preset highest gear, and an anti-corrosion treatment recommendation report is generated and sent to the management personnel. The anti-corrosion treatment recommendation report includes recommended agents and dosages.

[0013] To achieve the above-mentioned purpose, the second embodiment of the present invention proposes an artificial intelligence-based monument monitoring system, including: a data acquisition module, used to obtain multi-dimensional feature data of the monument to be monitored, the multi-dimensional feature data including mural surface image data, monument space environment parameters and supporting structure deformation data; a scheme planning module, used to generate an intelligent monitoring scheme according to the multi-dimensional feature data through a preset scheme planning model, the intelligent monitoring scheme including: at least one type of monitoring equipment and its layout position, layout method and working parameters; at least one alarm and its layout position, layout method, alarm strategy and working parameters; a monitoring process of the monitoring equipment; the scheme planning model is a machine learning model obtained by iterative training of historical monitoring scheme matching data; a monitoring deployment module, used to deploy monitoring equipment and alarms according to the intelligent monitoring scheme, and set working parameters; a monitoring execution module, which executes the monitoring process in the intelligent monitoring scheme, controls each monitoring device to coordinately monitor the monument, and enables the alarm to alarm according to the alarm strategy after abnormal data is monitored.

[0014] The artificial intelligence-based monument monitoring system proposed in the present invention also has the following additional technical features: According to one embodiment of the present invention, the above-mentioned artificial intelligence-based ancient monument monitoring system also includes: a real-time optimization module, which is used to obtain monitoring data in real time and when abnormal data appears in the monitoring data and meets the preset dynamic optimization conditions, the abnormal data and the monitoring data in the previous optimization cycle are packaged and input into the solution planning model to update the optimized intelligent monitoring solution.

[0015] According to one embodiment of the present invention, the training method of the scheme planning model includes the following steps: constructing a historical scheme library with weight annotations, the annotation dimensions include the disease detection efficiency weight, the equipment energy consumption weight and the data completeness weight, the weights of the three are set by the management personnel and the sum of the weights is 100%; using the Wasserstein GAN architecture for adversarial training, the generator input is multi-dimensional feature data, and the discriminator evaluates the distribution consistency of the output scheme and the expert scheme; adapting the pre-trained model to the new monument scene through transfer learning, and the number of fine-tuning layers does not exceed 20% of the total number of layers.

[0016] Beneficial effects of the present invention: 1. By obtaining multi-dimensional feature data of the monuments to be monitored and comprehensively reflecting the actual situation of the monuments to be monitored through multi-source data, the program planning model can efficiently and accurately determine the degradation and disease conditions of the monuments, and then plan intelligent monitoring programs in a targeted manner, build a full-dimensional monitoring system covering "surface-environment-structure", and realize closed-loop management from data collection to alarm response. It can monitor the monuments efficiently and accurately, realize efficient protection and maintenance of the monuments, and has good universality, which can meet the intelligent monitoring needs of murals and stone carvings in various regions; 2. Combine cultural relics protection standards, mechanical models, and the actual climate conditions of the sites where the monuments are located to set multi-dimensional dynamic optimization conditions, covering all risk dimensions of cultural relics protection; trigger scheme updates based on real-time data, optimize monitoring equipment parameters and alarm thresholds, form a "monitoring-optimization-feedback" closed loop, and achieve continuous dynamic evolution of monitoring schemes. Compared with traditional static monitoring schemes, it can greatly improve the speed of abnormal detection and reduce the false alarm rate; 3. In the training process of the scheme planning model, the innovative adversarial training with weighted annotation (WassersteinGAN) is adopted. The generator learns the multimodal feature distribution, and the discriminator constrains the generation quality through the expert scheme, which can better measure the distance between the generated distribution and the real distribution, so that the scheme generated by the generator is closer to the expert scheme in distribution, thereby improving the quality and rationality of the scheme; at the same time, through adversarial training, the generator can learn the data features in the historical scheme library, thereby generating diversified intelligent monitoring schemes to meet the needs of different monuments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a method flow chart of a method for monitoring ancient monuments based on artificial intelligence in an embodiment of the present invention; Figure 2 is a flow chart of a method for obtaining multi-dimensional feature data of a monument to be monitored in an embodiment of the present invention; Figure 3 is a method flow chart of a solution planning model training method in an embodiment of the present invention; Figure 4is a method flow chart of the corrosion risk early warning sub-process in an embodiment of the present invention; Figure 5 It is a system block diagram of a historical site monitoring system based on artificial intelligence in an embodiment of the present invention.

[0018] Explanation of the accompanying drawings: 1. Data acquisition module; 2. Solution planning module; 3. Monitoring deployment module; 4. Monitoring execution module; 5. Real-time optimization module. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The embodiment of the present invention discloses a method for monitoring ancient monuments based on artificial intelligence. Figure 1 , a monument monitoring method based on artificial intelligence, comprising the following steps: S1, obtaining multi-dimensional feature data of the monument to be monitored, the multi-dimensional feature data including mural surface image data, monument space environment parameters and supporting structure deformation data.

[0021] S2, generates an intelligent monitoring plan based on the multi-dimensional feature data through a preset plan planning model. The intelligent monitoring plan includes: at least one type of monitoring equipment and its layout location, layout method and working parameters; at least one alarm and its layout location, layout method, alarm strategy and working parameters; the monitoring process of the monitoring equipment. The plan planning model is a machine learning model obtained by iterative training through historical monitoring plan matching data.

[0022] The above monitoring equipment includes but is not limited to a three-dimensional laser scanning module (preferably with a scanning accuracy of 0.1mm, used to construct a high-precision digital model of the surface of the monument), a multispectral imaging module (preferably with a wavelength range covering 400-2500nm and a resolution better than 0.5nm), a macro camera (preferably equipped with a 200x optical zoom lens to identify surface microcracks), a video acquisition camera (preferably a fisheye panoramic lens to collect crowd flow videos), a non-contact vibration sensor, a millimeter-wave radar (used to monitor the approach distance of tourists), a fiber Bragg grating sensor (preferably buried in the base of the mural to monitor the strain sensitivity of ±1με), an environmental sensor, an infrared thermal imager, a particle counter (monitoring PM2.5 / PM10 concentration), an ultraviolet irradiance meter, an acoustic emission sensor (capturing elastic wave signals of crack expansion inside the mural), a gas composition analyzer, an air microbial sampler (air mold content), a biosensor (monitoring and identifying biological diseases of monuments) and a structural health monitoring system (including an inclinometer / accuracy of 0.001° and a crack meter / resolution of 0.01mm, real-time tracking of support structure deformation).

[0023] Among them, the alarm includes but is not limited to the directional sound wave alarm at the physical end (preferred frequency range: 16-18kHz), visual warning device, quick response protective cover and pre-set multi-level notification system (real-time alarm notification) and alarm log system (blockchain evidence storage) at the data end. Through the setting of monitoring equipment and alarms, high-precision equipment and blockchain evidence storage alarm system are used to build an integrated network of "perception-analysis-protection", and non-invasive deployment monitoring is realized according to the particularity of monument monitoring, such as fiber Bragg grating sensors buried in the base of murals, to achieve structural health monitoring while avoiding surface damage, and realize efficient and intelligent monitoring of monuments.

[0024] S3, deploy monitoring equipment and alarms according to the intelligent monitoring solution, and set working parameters.

[0025] S4, executes the monitoring process in the intelligent monitoring solution, controls various monitoring devices to conduct coordinated monitoring of the monuments, and activates the alarm according to the alarm strategy after abnormal data is monitored.

[0026] Therefore, by obtaining multi-dimensional characteristic data of the monuments to be monitored and comprehensively reflecting the actual situation of the monuments to be monitored through multi-source data, it is helpful for the program planning model to efficiently and accurately determine the degradation and disease conditions of the monuments, and then to plan intelligent monitoring programs in a targeted manner, build a full-dimensional monitoring system covering "surface-environment-structure", and realize closed-loop management from data collection to alarm response. It can monitor the monuments efficiently and accurately, realize efficient protection and maintenance of the monuments, and has good universality, which can meet the intelligent monitoring needs of murals and stone carvings in various regions.

[0027] In addition, the above steps also include S5, real-time acquisition of monitoring data and when abnormal data appears in the monitoring data that meets the preset dynamic optimization conditions, the abnormal data and the monitoring data in the previous optimization cycle are packaged and input into the solution planning model to update the optimized intelligent monitoring solution.

[0028] The dynamic optimization conditions are set by the management personnel. In this embodiment, the dynamic optimization conditions can trigger the scheme update when any of the following conditions is detected: The color difference of the surface of the ancient murals is ΔE>5 for 2 hours; refer to the CIELAB (International Illumination Commission) standard; Newly detected disease area ≥10mm²; Surface roughness change rate>15%; VOC (Volatile Organic Compounds) concentration exceeds the threshold (formaldehyde>0.08mg / m³, total sulfide>0.05mg / m³); The rate of change of temperature over time in environmental parameters |dT / dt|>3℃ / h or the rate of change of relative humidity over time |dRH / dt|>20% / h; Structural deformation acceleration>0.1mm / h²; The structural natural frequency changes>5%; Comparison with historical data shows that the annual deformation exceeds the baseline value by 30%.

[0029] Therefore, multi-dimensional dynamic optimization conditions are set up in combination with cultural relics protection standards, mechanical models, and actual climate conditions of the location of the ancient monuments, covering all risk dimensions of cultural relics protection; based on real-time data, the plan is updated to optimize monitoring equipment parameters and alarm thresholds, forming a "monitoring-optimization-feedback" closed loop, realizing continuous dynamic evolution of the monitoring plan, which can greatly improve the speed of anomaly detection and reduce the false alarm rate compared to static monitoring plans.

[0030] Reference Figure 2 The step S1 of obtaining the multi-dimensional feature data of the historic site to be monitored specifically includes the following steps: S11. Obtain millimeter-level precision 3D point cloud data of the monument surface through 3D laser scanning.

[0031] S12. Use a multispectral imaging device to collect spectral reflectance data of the mural surface, and fuse it with the three-dimensional point cloud data to generate mural surface image data.

[0032] The registration and fusion of point cloud and spectral data achieves spatiotemporal alignment and avoids misjudgment caused by spatial misalignment of multi-source data.

[0033] S13, deploy a micro-environment sensor array, continuously collect environmental parameters, and generate environmental parameters of the monument space. Environmental parameters include but are not limited to temperature, humidity, light intensity, air flow rate, and VOC concentration parameters.

[0034] S14. Retrieve historical monitoring data and restoration records of the monument through the historical data interface, and extract deformation data of the supporting structure.

[0035] Through the above steps, efficient and accurate collection of multi-source data of historical sites can be achieved. Multi-source data can comprehensively reflect the actual situation of the historical sites to be monitored, which will help the program planning model to efficiently and accurately determine the degradation and disease status of the historical sites, and provide a solid data foundation for subsequent intelligent decision-making.

[0036] Reference Figure 3 In one embodiment of the present invention, the training method of the scenario planning model includes the following steps: A1. Build a historical solution library: Build a historical solution library with weight annotation. The annotation dimensions include the disease detection efficiency weight, equipment energy consumption weight, and data completeness weight. The weights of the three are set by the management personnel and the sum of the weights is 100%. In this embodiment, the disease detection efficiency weight is preferably 30%, the equipment energy consumption weight is preferably 25%, and the data completeness weight is preferably 45; A2. Adversarial training: The Wasserstein GAN architecture is used for adversarial training. The generator input is multi-dimensional feature data, and the discriminator evaluates the distribution consistency between the output solution and the expert solution. A3. Transfer learning: The pre-trained model is adapted to the new historic site scenario through transfer learning, and the number of fine-tuning layers does not exceed 20% of the total number of layers.

[0037] Through the above steps, the historical solution library introduces three-dimensional weights of disease detection efficiency, energy consumption, and data completeness, taking into account both protection effect and economy, and can greatly improve the comprehensive effectiveness of decision-making and planning solutions in practical applications; innovatively adopts adversarial training with weighted annotation (Wasserstein GAN), the generator learns multimodal feature distribution, and the discriminator constrains the generation quality through expert solutions, which can better measure the distance between the generated distribution and the true distribution, so that the solution generated by the generator is closer to the expert solution in distribution, improving the quality and rationality of the solution; at the same time, through adversarial training, the generator can learn the data features in the historical solution library, thereby generating diversified intelligent monitoring solutions to meet the needs of different monuments; through transfer learning, the knowledge and features learned by the machine learning model in other monument scenes can be used to quickly adapt to new monument scenes, reducing the training time and the demand for a large amount of labeled data in new scenes, and at the same time avoiding overfitting by limiting the number of fine-tuning layers to ≤20%.

[0038] Reference Figure 4Since the biggest risk faced in the protection of murals and stone carvings is corrosion from the outside, corrosion risk warning is necessary for the intelligent monitoring of various types of monuments. The monitoring process includes the corrosion risk warning sub-process, which specifically includes the following steps: B1. Calculate the corrosion index CI based on the gas composition in the monitored environmental parameters: CI = 0.35 × [SO 2 ]+0.28×[NO 2 ]+0.37×[Cl - ]; Where CI is the corrosion index, [SO 2 ] is the concentration of sulfur dioxide in the air, [NO 2 ] is the concentration of nitrogen dioxide in the air, [Cl - ] is the chloride ion concentration in the air; B2. Obtain the humidity data RH in the environmental parameters. When CI>0.85 and RH>65%, determine whether there is a quick response protective cover in the area. If so, activate the quick response protective cover, increase the gas sampling frequency to the preset highest gear, and generate an anti-corrosion treatment recommendation report and send it to the management personnel. The anti-corrosion treatment recommendation report includes recommended agents and dosages.

[0039] Therefore, by breaking through the single parameter alarm mode and realizing accurate early warning through multi-factor joint calculation (CI×humidity), the corrosion event detection time can be greatly advanced, and the gas sampling frequency can be automatically increased after the corrosion risk is discovered, which can ensure the timeliness of the data and further improve the timeliness and accuracy of the corrosion risk warning.

[0040] The embodiment of the present invention also discloses a monitoring system based on artificial intelligence. Figure 5 , an artificial intelligence-based monitoring system, comprising: Data acquisition module 1, used to acquire multi-dimensional feature data of the monument to be monitored, the multi-dimensional feature data including mural surface image data, monument space environment parameters and supporting structure deformation data; The solution planning module 2 is used to generate an intelligent monitoring solution through a preset solution planning model according to the multi-dimensional feature data. The intelligent monitoring solution includes: at least one type of monitoring equipment and its layout location, layout method and working parameters; at least one alarm and its layout location, layout method, alarm strategy and working parameters; the monitoring process of the monitoring equipment. The solution planning model is a machine learning model obtained by iterative training through historical monitoring solution matching data; Monitoring deployment module 3, used to deploy monitoring equipment and alarms according to the intelligent monitoring solution, and set working parameters; Monitoring execution module 4 executes the monitoring process in the intelligent monitoring solution, controls various monitoring devices to monitor the monuments in a coordinated manner, and activates the alarm according to the alarm strategy after abnormal data is monitored. By obtaining multi-dimensional feature data of the monuments to be monitored, the actual situation of the monuments to be monitored can be fully reflected through multi-source data, which helps the program planning model to efficiently and accurately determine the degradation and disease conditions of the monuments, and then plan intelligent monitoring solutions in a targeted manner, build a full-dimensional monitoring system covering "surface-environment-structure", and realize closed-loop management from data collection to alarm response. It can monitor the monuments efficiently and accurately, realize efficient protection and maintenance of the monuments, and has good universality, which can meet the intelligent monitoring needs of murals and stone carvings in various regions.

[0041] In addition, refer to Figure 5 , a monitoring system based on artificial intelligence also includes: a real-time optimization module 5, which is used to obtain monitoring data in real time and when abnormal data appears in the monitoring data and meets the preset dynamic optimization conditions, the abnormal data and the monitoring data in the previous optimization cycle are packaged and input into the program planning model to update and optimize the intelligent monitoring program. Combined with the cultural relics protection standards, mechanical models, and the actual climate conditions of the site of the monument, multi-dimensional dynamic optimization conditions are set to cover the full risk dimension of cultural relics protection; based on real-time data, the program is updated, the monitoring equipment parameters and alarm thresholds are optimized, and a "monitoring-optimization-feedback" closed loop is formed to achieve continuous dynamic evolution of the monitoring program. Compared with the traditional static monitoring program, it can greatly improve the speed of abnormal detection and reduce the false alarm rate.

[0042] The training method of the scheme planning model includes the following steps: constructing a historical scheme library with weighted annotations, where the annotation dimensions include the weight of disease detection efficiency, the weight of equipment energy consumption, and the weight of data completeness. The weights of the three are set by managers and the total weight is 100%; using the Wasserstein GAN architecture for adversarial training, the generator input is multi-dimensional feature data, and the discriminator evaluates the distribution consistency of the output scheme and the expert scheme; adapting the pre-trained model to the new monument scene through transfer learning, and the number of fine-tuning layers does not exceed 20% of the total number of layers.

[0043] An embodiment of the present invention further discloses a computer-readable storage medium, which stores a computer program that can be loaded by a processor and executed as in the above method. The computer-readable storage medium includes, for example: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0044] In summary, the present invention obtains multi-dimensional feature data of the monuments to be monitored, and comprehensively reflects the actual situation of the monuments to be monitored through multi-source data, which helps the program planning model to efficiently and accurately determine the degradation and disease conditions of the monuments, and then plans intelligent monitoring programs in a targeted manner, builds a full-dimensional monitoring system covering "surface-environment-structure", and realizes closed-loop management from data acquisition to alarm response. It can monitor the monuments efficiently and accurately, realize efficient protection and maintenance of the monuments, and has good universality, which can meet the intelligent monitoring needs of murals and stone carvings in various regions; the present invention combines the cultural relics protection standards, mechanical models, and the actual climate conditions of the monuments to set multi-dimensional dynamic optimization conditions, covering the full risk dimension of cultural relics protection; based on real-time data triggering program updates, optimizes monitoring equipment parameters and alarm thresholds, forms a "monitoring-optimization-feedback" closed loop, and realizes continuous dynamic evolution of monitoring programs. Compared with traditional static monitoring programs, it can greatly improve the abnormality detection speed and reduce the false alarm rate; in the training process of the program planning model, the present invention innovatively adopts adversarial training with weight annotation (Wasserstein GAN), the generator learns the multimodal feature distribution, and the discriminator constrains the generation quality through expert solutions, which can better measure the distance between the generated distribution and the true distribution, making the solutions generated by the generator closer to the expert solutions in distribution, improving the quality and rationality of the solutions; at the same time, through adversarial training, the generator can learn the data features in the historical solution library, thereby generating diversified intelligent monitoring solutions to meet the needs of different monuments.

[0045] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0046] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0047] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present invention belong.

[0048] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0049] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system.

[0050] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0051] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0052] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for monitoring ancient monuments based on artificial intelligence, characterized in that: The following steps are involved: S1, obtaining multi-dimensional feature data of the monument to be monitored, the multi-dimensional feature data including mural surface image data, monument space environment parameters and supporting structure deformation data; S2, generating an intelligent monitoring solution through a preset solution planning model according to the multi-dimensional feature data, wherein the intelligent monitoring solution includes: at least one type of monitoring equipment and its layout location, layout method and working parameters; at least one alarm and its layout location, layout method, alarm strategy and working parameters; and a monitoring process of the monitoring equipment, wherein the solution planning model is a machine learning model obtained by iterative training through historical monitoring solution matching data; S3, deploy monitoring equipment and alarms according to the intelligent monitoring solution, and set working parameters; S4, executes the monitoring process in the intelligent monitoring solution, controls various monitoring devices to conduct coordinated monitoring of the monuments, and activates the alarm according to the alarm strategy after abnormal data is monitored.

2. The artificial intelligence-based monument monitoring method according to claim 1 is characterized in that: The monitoring equipment includes: a three-dimensional laser scanning module, a multi-spectral imaging module, a macro camera, a video acquisition camera, a non-contact vibration sensor, a millimeter wave radar, a fiber Bragg grating sensor, an environmental sensor, an infrared thermal imager, a particle counter, an ultraviolet irradiance meter, an acoustic emission sensor, a gas composition analyzer, an air microbial sampler, a biosensor and a structural health monitoring system.

3. The artificial intelligence-based monument monitoring method according to claim 1 is characterized in that: The alarm includes: a directional sound wave alarm, a visual warning device and a quick response protective cover at the physical end; a pre-set multi-level notification system and an alarm log system at the data end.

4. The method for monitoring ancient monuments based on artificial intelligence according to claim 1, characterized in that: The step S1 of obtaining the multi-dimensional feature data of the historic site to be monitored specifically includes the following steps: S11. Obtain millimeter-level precision 3D point cloud data of the monument surface through 3D laser scanning; S12, using a multi-spectral imaging device to collect spectral reflectance data of the mural surface, and merging the spectral reflectance data with the three-dimensional point cloud data to generate image data of the mural surface; S13, deploying a micro-environment sensor array to continuously collect environmental parameters and generate environmental parameters of the historic space; the environmental parameters include temperature, humidity, light intensity, air flow rate and VOC concentration parameters; S14. Retrieve historical monitoring data and restoration records of the monument through the historical data interface, and extract deformation data of the supporting structure.

5. The artificial intelligence-based monument monitoring method according to claim 1 is characterized in that: Also includes: S5, real-time acquisition of monitoring data, and when abnormal data appears in the monitoring data that meets the preset dynamic optimization conditions, the abnormal data and the monitoring data in the previous optimization cycle are packaged and input into the solution planning model to update the optimized intelligent monitoring solution.

6. The artificial intelligence-based monument monitoring method according to claim 1 is characterized in that: The training method of the scenario planning model comprises the following steps: Construct a historical solution library with weight annotation. The annotation dimensions include disease detection efficiency weight, equipment energy consumption weight, and data completeness weight. The weights of the three are set by the management personnel and the total weight is 100%; The Wasserstein GAN architecture is used for adversarial training. The generator input is multi-dimensional feature data, and the discriminator evaluates the distribution consistency between the output solution and the expert solution. The pre-trained model is adapted to the new historic site scenario through transfer learning, and the number of fine-tuning layers does not exceed 20% of the total number of layers.

7. The artificial intelligence-based monument monitoring method according to claim 1 is characterized in that: The monitoring process includes a corrosion risk early warning sub-process, and the corrosion risk early warning sub-process specifically includes the following steps: The corrosion index CI is calculated based on the gas composition in the monitored environmental parameters. CI = 0.35 × [SO2] + 0.28 × [NO2] + 0.37 × [Cl - ], where CI is the corrosion index, [SO2] is the concentration of sulfur dioxide in the air, [NO2] is the concentration of nitrogen dioxide in the air, [Cl - ] is the concentration of chloride ions in the air; Obtain the humidity data RH in the environmental parameters. When CI>0.85 and RH>65%, determine whether there is a quick response protective cover in the area. If so, start the quick response protective cover, increase the gas sampling frequency to the preset highest gear, and generate an anti-corrosion treatment recommendation report and send it to the management personnel. The anti-corrosion treatment recommendation report includes recommended agents and dosages.

8. An artificial intelligence-based monument monitoring system, characterized in that: include: A data acquisition module is used to acquire multi-dimensional feature data of the monument to be monitored, wherein the multi-dimensional feature data includes mural surface image data, monument space environment parameters, and supporting structure deformation data; A scheme planning module is used to generate an intelligent monitoring scheme according to the multi-dimensional feature data through a preset scheme planning model, wherein the intelligent monitoring scheme includes: at least one type of monitoring equipment and its layout location, layout method and working parameters; at least one alarm and its layout location, layout method, alarm strategy and working parameters; and a monitoring process of the monitoring equipment. The scheme planning model is a machine learning model obtained by iterative training through historical monitoring scheme matching data. Monitoring deployment module, used to deploy monitoring equipment and alarms according to the intelligent monitoring solution, and set working parameters; The monitoring execution module executes the monitoring process in the intelligent monitoring solution, controls various monitoring devices to conduct collaborative monitoring of the historical sites, and activates the alarm according to the alarm strategy after abnormal data is monitored.

9. The artificial intelligence-based monument monitoring system according to claim 8 is characterized in that: Also includes: The real-time optimization module is used to obtain monitoring data in real time and when abnormal data appears in the monitoring data and meets the preset dynamic optimization conditions, the abnormal data and the monitoring data in the previous optimization cycle are packaged and input into the solution planning model to update the optimized intelligent monitoring solution.

10. The artificial intelligence-based monument monitoring system according to claim 8, characterized in that: The training method of the scenario planning model comprises the following steps: Construct a historical solution library with weight annotation. The annotation dimensions include disease detection efficiency weight, equipment energy consumption weight, and data completeness weight. The weights of the three are set by the management personnel and the total weight is 100%; The Wasserstein GAN architecture is used for adversarial training. The generator input is multi-dimensional feature data, and the discriminator evaluates the distribution consistency between the output solution and the expert solution. The pre-trained model is adapted to the new historic site scenario through transfer learning, and the number of fine-tuning layers does not exceed 20% of the total number of layers.

Images