Real-time environmental data linkage explanation control method and system

Abstract

The invention discloses a real-time environment data linkage explanation control method and system, and belongs to the technical field of environment intelligent explanation. The real-time environmental data linkage explanation control method comprises the following steps: determining corresponding species information by combining field data acquired in a target area with a deep learning model; environment factors are extracted, a knowledge graph is called to carry out correlation analysis on the species information and the environment factors, and an explanation object is determined; recording the field data corresponding to the explanation object as a marking factor, and associating the marking factor with an explanation database to obtain an explanation entry; and inputting the explanation entry into the large model to generate explanation content. According to the method, personalized explanation containing real-time environment parameters can be generated, and personalized explanation content is generated through dynamic data linkage.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent environmental interpretation, and in particular to a real-time environmental data linkage interpretation control method and system. Background Art

[0002] Providing accurate and comprehensive nature interpretation content to visitors and researchers in nature reserves, national parks, and other areas is crucial for ecological conservation education and scientific research. Current research in nature interpretation technology is focusing on intelligent services based on the Internet of Things, artificial intelligence, and multimodal interaction. These technologies aim to generate personalized interpretation content through real-time environmental data (such as infrared sensors monitoring animal activity), positioning technologies (such as UWB and satellite tracking), and dynamic algorithms.

[0003] For example, patent publication number CN101018062A, entitled "Intelligent Tour Guide System and Its Operating Method," discloses an intelligent wireless positioning transmitter installed in areas of tourist attractions requiring tours. It features a built-in wireless code generator for transmitting coded information. An infrared transmitter, installed at appropriate locations within the tour guide's interior, transmits infrared signals. The electronic tour guide terminal, equipped with a wireless signal receiving module, an infrared signal receiving module, a display module, a voice module, a single-chip microcomputer, memory, and a data exchange interface, receives the coded information to identify routes and provide intelligent guidance. The management and statistics system, including statistical algorithms, unlocking procedures, and a database system, ensures traffic statistics and equipment management within the scenic area. While visitors can freely choose their tour routes, images and audio are fully synchronized, providing personalized services. The solution also boasts low cost, compact size, light weight, and simple operation. However, the pre-set and static content of the tour guides results in repeated basic explanations, lacking real-time customization based on environmental data.

[0004] For example, the invention patent with publication number CN119342295A and invention name "Panoramic interactive display method and system based on AI multimodal fusion" discloses that by integrating the visual and audio information of the current panoramic material and combining multiple panoramic interactive hotspot information, efficient and accurate multimodal fusion is achieved. It not only generates a panoramic interaction model with rich interactivity, but also further extracts the hotspot interaction space features. Although it can automatically adjust the viewing angle according to user needs or preset scenarios and trigger corresponding commentary audio or interactive animation, significantly improving the immersion and interactivity of the panoramic display and optimizing the user experience, it relies on preset panoramic materials and static hotspot triggering mechanism, and cannot achieve real-time environmental data fusion and dynamic behavior response, resulting in significant deficiencies in its adaptability to dynamic scenes.

[0005] Traditional interpretation methods often rely on manually written, fixed content, and are unable to provide flexible and personalized interpretation based on real-time on-site data and different environmental conditions. With the development of information technology, the use of modern technology to achieve automated and intelligent generation of natural interpretation content has become a trend.

[0006] It can be seen that there is an urgent need to develop a new real-time environmental data linkage interpretation control method to solve the current defects and shortcomings. Summary of the Invention

[0007] In view of this, the main purpose of the present invention is to provide a real-time environmental data linkage interpretation control method and system, in order to at least partially solve the above technical problems.

[0008] To achieve the above objectives, as a first aspect of the present invention, a real-time environmental data linked commentary control method is proposed, comprising:

[0009] Field data acquired in the target area is combined with deep learning models to determine the corresponding species information;

[0010] Extract environmental factors, call the knowledge graph to conduct correlation analysis between species information and environmental factors, and determine the interpretation object;

[0011] Record the field data corresponding to the interpretation object as a marking factor, and associate the marking factor with the interpretation database to obtain the interpretation entry;

[0012] Input the explanation entries into the large model to generate the explanation content.

[0013] As a second aspect of the present invention, a real-time environmental data linked interpretation control system is also proposed, comprising:

[0014] Data collection units, including infrared cameras, drone cameras, or fixed surveillance cameras for collecting image data, microphone arrays, bird song identifiers, or underwater sonar for collecting sound data, and weather stations, geographic information systems, and air quality monitors for obtaining abiotic environmental data;

[0015] The species identification unit uses a deep learning model trained with species samples to receive data from the data acquisition unit, identify and determine species information, and output the results to the environmental analysis unit;

[0016] The environmental analysis unit is used to extract environmental factors, call the knowledge graph to perform correlation analysis between species information and environmental factors, and determine the interpretation object and transmit it to the interpretation association unit;

[0017] An interpretation association unit is used to record the field data corresponding to the interpretation object as a marking factor, associate it with the interpretation database to obtain an interpretation entry, and send the interpretation entry to the content generation unit; and

[0018] The content generation unit is used to input interpretation items into the large model to generate interpretation content, optimize it in combination with field data and historical data, and output the generated content.

[0019] As a third aspect of the present invention, an electronic device is further provided, comprising:

[0020] a memory for storing computer programs executable on the processor;

[0021] The processor is used to execute the computer program stored in the memory to implement the real-time environmental data linkage interpretation control method as described above.

[0022] As a fourth aspect of the present invention, a processor executable program is also proposed. The processor executable program can be executed directly by the processor or can be executed by the processor after compilation, and is used to execute the real-time environmental data linkage interpretation control method as described above.

[0023] As a fifth aspect of the present invention, a storage medium is also proposed, which stores non-volatile program code that is executable by a processor or executable after compilation, and the program code is used to execute the real-time environmental data linkage interpretation control method as described above.

[0024] Based on the above technical solutions, the real-time environmental data linkage interpretation control method and system of the present invention have at least one of the following beneficial effects compared to the prior art:

[0025] 1. By collecting images, sounds, and environmental data through infrared cameras, microphone arrays, and other devices, and combining convolutional neural networks (CNNs), bidirectional LSTM networks, and Bayesian voting algorithms, multimodal data complementation is achieved, improving the accuracy and robustness of species identification.

[0026] 2. By constructing a knowledge graph of species-environmental factors and using Spearman correlation analysis, canonical correspondence analysis (CCA) and fuzzy logic algorithms, we can explore the deep connection between species and the environment, greatly improving the scientific nature of the interpretation object determination; by using fuzzy logic to deal with the fuzzy boundaries of environmental parameters, the trigger conditions are more in line with the real ecological scene, and the matching accuracy of interpretation items is improved compared with the traditional threshold method.

[0027] 3. By associating the interpretation database with marker factors, combining large models and dynamic parameter injection, personalized interpretations containing real-time environmental parameters are generated. Dynamic data is linked to generate personalized interpretation content, and the monitoring point trigger mechanism is used to solve the interpretation lag problem of traditional guide systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0029] Figure 1 It is a flow chart of the real-time environmental data linkage interpretation control method of the present invention;

[0030] Figure 2 is a schematic diagram of an electronic device for the real-time environmental data linkage interpretation control method of the present invention;

[0031] In the above drawings, the meanings of the reference numerals are as follows:

[0032] 410-Processor 420-Communication Interface

[0033] 430-Memory 440-Communication Bus DETAILED DESCRIPTION

[0034] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0035] The terms used in the present invention are only for the purpose of describing specific embodiments and are not intended to limit the embodiments of the present invention. The singular forms "a", "an", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise.

[0036] The inventors found that the traditional interpretation methods in the prior art often rely on manually written fixed content, which cannot provide flexible and personalized interpretation based on real-time data on site and different environmental conditions. After in-depth research, they found that the solution to generate personalized natural interpretation content can be solved by combining species information and environmental data for comprehensive processing. Figure 1 As shown, the inventors have proposed a real-time environmental data linkage interpretation control method, comprising:

[0037] Field data acquired in the target area is combined with deep learning models to determine the corresponding species information;

[0038] Extract environmental factors, call the knowledge graph to conduct correlation analysis between species information and environmental factors, and determine the interpretation object;

[0039] Record the field data corresponding to the interpretation object as a marking factor, and associate the marking factor with the interpretation database to obtain the interpretation entry;

[0040] Input the explanation entries into the large model to generate the explanation content.

[0041] Field data includes image data, sound data and non-biological environmental data. Image data is collected through infrared cameras, drone cameras or fixed monitoring cameras deployed in the target area. Sound data is collected through microphone arrays, bird song identifiers or underwater sonar. Non-biological environmental data is obtained in real time through meteorological stations, geographic information systems and air quality monitors.

[0042] Species information includes species names and the number of species associated with the species names, morphological characteristics, species behavior and location information. The number of species is determined by matching the target detection model or tracking algorithm, the morphological characteristics are determined by feature matching through ResNet combined with the knowledge graph, the species behavior is determined by the time series model combined with the knowledge graph mapping behavior label, and the location information is determined by GPS positioning.

[0043] Among them, target detection models (such as YOLO and FasterR-CNN) are used for image data to identify species bounding boxes and remove duplicates, that is, non-maximum suppression (NMS) is used to count the number of individuals in a single frame image; tracking algorithms (such as DeepSORT) are used for video streams to count the number of species within a certain period of time through feature matching of consecutive frames (such as appearance features and motion trajectories).

[0044] The present invention will be further described below through specific examples. It should be noted that the following examples are merely illustrative and are not intended to limit the present invention. Based on the embodiments of the present invention shown below, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the embodiments of the present invention.

[0045] Example 1

[0046] In this embodiment, further, taking the Himalayan marmot in the Liancheng National Nature Reserve in Gansu as an example, the infrared camera deployed in the reserve captured a single-frame image containing a Himalayan marmot with a resolution of 1920×1080, and the marmots were located in the center and right of the picture (3 in total). The YOLOv8 model (pre-trained weights based on the COCO dataset, with 5,000 new Himalayan marmot annotated data) was used to input the image and output 3 bounding boxes with confidence levels of 0.95, 0.92, and 0.88, corresponding to marmot individuals A, B, and C. The IoU threshold was set to 0.5, and the 3 bounding boxes had no overlap (IoU < 0.3), so all were retained and the number of single frames was recorded as 3. In the video stream (30fps, lasting 5 minutes) obtained by the drone camera, 3 marmots (ID = 1, 2, 3) were detected in the 1st frame, and a new marmot (ID = 4, entering from the left side of the picture) appeared in the 100th frame. The DeepSORT algorithm was used to extract each marmot's fur texture (dark brown back, light yellow abdomen) and body proportions (body length 50-60 cm) as appearance feature vectors. The marmot with ID = 1 moved toward the right side of the frame in frames 50-150, with a continuous trajectory. The marmot with ID = 4 maintained an independent trajectory in frames 100-300, without overlapping with other IDs. A total of four unique IDs were detected within 5 minutes. Combined with the single-frame detection results, it was determined that four Himalayan marmots were currently present in the area.

[0047] The morphological characteristics of the Himalayan marmot were determined by using ResNet combined with a knowledge graph for feature matching. First, a clear frontal image of the marmot was selected, resized to 224×224, and fed into a ResNet50 model (freezing the first four layers and training the last three layers). Feature extraction yielded shallow features such as ear shape (short, rounded ears with slightly blunt tips) and nose features (black nose with darker fur surrounding the nose), as well as deeper features such as hair texture (brown hair on the back with black tips, lighter hair on the abdomen) and body structure (sturdy body with a short, fluffy tail). The knowledge graph was used to match the morphological attributes of the "Himalayan marmot" node: body length: 50-65 cm, tail length: 10-15 cm (1 / 5-1 / 4 of body length); coat color: dark brown on the back, light yellow on the abdomen, with light gray stripes on the sides; and skull features: robust cheekbones and orange incisors. By verifying that the "short round ears" and "short fluffy tail" extracted by ResNet are consistent with the description in the knowledge base, the standardized morphological information is output: the body is stout, the body length is 55±5cm, and the tail length is about 12cm; the hair on the back is dark brown with black tips, the abdomen is light yellow, and there are light gray vertical stripes on the sides of the body; the ears are short and round, the nose is black, and the incisors are exposed in orange.

[0048] A temporal model was used to identify the behavior of Himalayan marmots. 100 frames of continuous video (resolution 1280×720, frame rate 25fps) of marmot activity were selected, and the bounding box coordinates and appearance features of each frame were extracted. Analysis of the marmot's trajectory revealed that from frames 1 to 50, the marmot with ID 1 was moving back and forth between its burrow and the vegetation area (straight-line distance 10 meters, frequency 3 times / minute). From frames 60 to 80, the limb movements detected were those of its front paws digging (joint angle change >30°, duration 5 frames), consistent with "digging" behavior. The behavioral labels from the "Himalayan Marmot Behavioral Research" were associated: moving back and forth between the vegetation area and the mouth opening and closing frequency >10 times / minute were labeled as foraging behavior; continuous digging with the front paws, a forward body tilt angle >45°, and duration >10 seconds were labeled as digging behavior. Combining the motion trajectory and limb movements output by the time series model, the foraging behavior labels (92% confidence) and digging behavior labels (88% confidence) were matched, and the species behavior was finally output: two Himalayan marmots were currently detected foraging within 20 meters of the cave (nibbling on Stipa and Kobresia), and one marmot was digging and expanding its nest at the entrance of the cave. The limb movements were manifested as rapid digging of the soil with the front paws and periodic forward leaning of the body.

[0049] The infrared camera was installed at an altitude of 2800 meters, with GPS coordinates recorded: (N36°52′18″, E102°45′30″). The camera was facing due south, with a field of view covering an area 50 meters in front. The Himalayan marmot's pixel coordinates in the image were (500, 600) and (1200, 400). ID = 1: 15 meters from the camera, azimuth 10° (east); ID = 2: 25 meters from the camera, azimuth 350° (west). The GPS coordinates are combined with azimuth and distance using a latitude and longitude conversion formula (such as the Haversine formula) to output location information: ID = 1: N36°52′18.5″, E102°45′31.2″, 2800 m above sea level (located in the transition zone between coniferous forest and meadow, 5 meters from the nearest cave); ID = 2: N36°52′17.8″, E102°45′28.9″, 2798 m above sea level (located in an exposed rocky area, surrounded by piles of marmot feces).

[0050] For vocal species such as birds, the voiceprint recognition algorithm, specifically the GMM-HMM algorithm, can be used to distinguish the calls of different individuals, combined with sound source localization technology (microphone array) to estimate the number of species in the area.

[0051] This example takes the golden eagle in the Liancheng National Nature Reserve in Gansu Province as an example. 2) deployed a 7-element uniform circular microphone array (50m radius), with each microphone spaced 10m apart, to synchronously capture sound signals at 16kHz and 16-bit resolution, covering the typical golden eagle call frequency range (0.5-8kHz). A GPS module was installed at the center of the array to record the coordinates (N36°50′20″, E102°48′30″), and a synchronized clock module was used to ensure that the time error between microphones was less than 1μs. Golden eagle calls were recorded in various scenarios (such as territorial announcements, mate calls, and chicks begging for food), resulting in a total of 2,000 valid audio segments (each 2-5 seconds long) and individual IDs (A, B, C, confirmed by banding). Spectral subtraction was used to remove background noise (such as wind and stream sounds), retaining signals with a signal-to-noise ratio ≥15dB. The audio segments were segmented into 50ms frame lengths with 25ms overlap, and Mel-frequency cepstral coefficients (MFCCs) (20-dimensional) and first-order difference features (ΔMFCCs) were extracted to form a 40-dimensional feature vector.

[0052] For each golden eagle's vocal characteristics (such as the "high-frequency scream + low-frequency trill" pattern of individual A), a 5-component GMM was used to fit its characteristic distribution: where ω i is the mixing weight, μ i and ∑ i is the mean and covariance of the i-th Gaussian component.

[0053] Based on the temporal characteristics of the golden eagle's call (such as the three-state sequence of "scream-pause-trill"), a three-state left-to-right HMM is established, and each state corresponds to the GMM distribution:

[0054] State 1: high-frequency screaming (MFCC mean: [20, 18, ...], energy > 80dB);

[0055] State 2: silent pause (energy < 40dB, lasting 0.5-1 second);

[0056] State 3: low-frequency vibrato (MFCC mean: [15, 22, ...], frequency < 2kHz).

[0057] The Baum-Welch algorithm is used to train the model parameters to maximize the log-likelihood of the training data. This is to find the model parameters that best explain the observed data through iterative optimization when the implicit state is unknown.

[0058] 500 calls that did not participate in the training were tested, and the individual recognition accuracy was >95%. Misjudgments mainly occurred in individuals of the same family (call similarity >85%), which were eliminated through subsequent GPS positioning.

[0059] After preprocessing, the real-time collected chirping signals are input into all trained HMM models, the log likelihood is calculated (for example, the model output of individual A has a log likelihood of -200, and the model output of individual B has a log likelihood of -250), and the individual ID corresponding to the maximum value is selected.

[0060] If all model likelihoods are < -300 (the threshold is determined by the validation set), it is determined to be a new individual (ID automatically increments, such as D, E).

[0061] A microphone array is used for sound source localization. The time difference between the sound reaching different microphones is calculated (for example, microphone 1 receives the signal 0.5ms earlier than microphone 2). The sound source coordinates (x, y, z) are solved using the least squares method based on the time difference and the geometry of the microphone array. The positioning accuracy is less than 5m (the vertical error is less than 10m, considering the golden eagle's flight altitude).

[0062] If the distance between the positioning coordinates of two call signals is less than 50m and the likelihood difference is less than 10, they are determined to be the same individual (to avoid ID duplication caused by flight movement);

[0063] If the positioning coordinate distance is greater than 100m and the likelihood difference is greater than 50, they are determined to be different individuals (e.g., individual A is on the east ridge and individual B is in the west valley).

[0064] Continuous monitoring was performed for 24 hours, and the individual ID, location coordinates, and timestamp of each valid call were recorded. The quantitative statistical rule was that repeated calls of the same ID within 10 minutes were counted only once, and repeated counting was avoided when the golden eagle called at a high frequency. The location coordinates were clustered using DBSCAN (radius 50m, minimum number of points 1), and each cluster represented an individual activity area.

[0065] Combined with drone visible light / infrared image monitoring (to identify the body shape and wingspan characteristics of golden eagles), the voiceprint recognition results were verified: the voiceprint detected three individuals (ID-A, B, and C), and the drone simultaneously observed three golden eagles active in different areas, with the same number; a certain rainfall caused the voiceprint detection to miss a young bird, which was supplemented by image recognition, and the final corrected number was 4.

[0066] The estimated number of species is: 2 In the monitoring area, through 72 hours of continuous monitoring, a total of 4 golden eagles (2 adults + 2 sub-adults) were detected.

[0067] Species information is determined through the following steps: After preprocessing the image data and sound data, a convolutional neural network model is used to extract image features, and a bidirectional LSTM network is used to extract sound features;

[0068] Bayesian voting is performed on the output results of the convolutional neural network and the bidirectional LSTM network, and the fusion rules are set as follows: if the confidence in the convolutional neural network and the bidirectional LSTM network is greater than the species recognition threshold and the species name is consistent, the species name is output; the confidence of the convolutional neural network and the bidirectional LSTM network is sorted and the joint probability is determined, and the species name corresponding to the maximum joint probability is output; if the joint probabilities are all less than the joint lower limit threshold, the no data result is output.

[0069] Using the Himalayan marmot as an example, the fusion rule is as follows: If the image is identified as "Himalayan marmot" (confidence ≥ 80%) and no other species characteristics are detected in the sound, the output is "Himalayan marmot." If the image confidence (70%-80%) and the sound characteristics (such as the typical warning call, which is a monosyllabic "chirp" pulse) match, the final confidence is increased to 90%. When the drone captures the blurred outline of an animal in the plateau area (image confidence 75%), combined with the Himalayan marmot call detected by the sonar at the same time, the species is determined to be a Himalayan marmot after fusion.

[0070] The output results of the convolutional neural network and the bidirectional LSTM network are verified for spatiotemporal consistency. The azimuth of the output sound source is estimated through the time delay of the microphone array and cross-validated with the location information of the image data to detect whether the image data and sound data are both within the target area and eliminate cross-regional interference, such as the incorrect association between distant sounds and nearby images.

[0071] Environmental factors were obtained from field data within the target area. These factors include altitude, slope, aspect, and the intensity of human disturbance. The intensity of human disturbance experienced by each vegetation type was measured by measuring the area within different distance intervals from the source of human disturbance. Slope is a key topographic factor in vegetation distribution, affecting factors such as light intensity, duration of daylight, temperature, soil moisture, and soil physical and chemical properties. Furthermore, vegetation on different slopes experiences varying degrees of disturbance. Natural vegetation within the reserve is primarily distributed on steep and abrupt slopes with gradients of 15 to 35°. Deciduous broad-leaved forests and deciduous broad-leaved shrubs predominate on flat and gentle slopes with gradients of 0 to 15°. Cold-temperate coniferous forests, temperate coniferous forests, and deciduous broad-leaved shrubs predominate on steep slopes with gradients greater than 35°. The knowledge graph was used to conduct a correlation analysis between species information and environmental factors. Specifically, a species-environmental factor matrix was constructed. The correlations between species information and environmental factors were determined through Spearman correlation analysis and / or canonical correspondence analysis. Based on these correlations, a species distribution prediction model was established. This model was validated using a Monte Carlo permutation test, with the total explanatory power of environmental factors exceeding 30%. The results showed that altitude and canopy density had the most significant impact on species distribution, while distance from disturbance sources had the least significant influence.

[0072] The process of associating the marking factors with the interpretation database to obtain interpretation entries is as follows: first, the interpretation entries of a single marking factor are screened using a hash index algorithm, and then the interpretation entries corresponding to multiple marking factors are processed using a fuzzy logic algorithm.

[0073] The terrain of Liancheng Nature Reserve is characterized by high in the west and low in the east, with an altitude between 1870 and 3616 meters. The wild plants distributed there, such as Peach Blossom Seven, Sichuan Red Peony Root, Feather-leaved Lilac and Two-leaved Cymbidium, are the habitats of golden eagles. Among them, Feather-leaved Lilac belongs to the lilac shrub, with an altitude of 2200 to 2400 meters.

[0074] The field data obtained are as follows: Species name: Golden Eagle; Number of species: 3 (2 adults, 1 subadult); Morphological characteristics: Wingspan 2.2-2.3m, dark brown feathers on the back, white base of tail feathers; Species behavior: territorial declaration calls, high-altitude gliding, and catching hares with claws; Location information: N36°50′20″, E102°48′30″, 2850m above sea level (located in the concentrated distribution area of ​​Syringa chinensis, with a vegetation coverage rate of 70%).

[0075] The environmental factors obtained are: altitude: 2850m; vegetation type: Syringa chinensis; temperature: 18℃; light intensity: 8000lux; wind speed: 5m / s.

[0076] A unique key value is generated for the species name of the golden eagle through a hash algorithm, directly locating the basic entries of the golden eagle (species classification, habitat, endangered status, legends, ecological value, and evolutionary history);

[0077] A B-tree index was used to search for the range (2000-3500m) above sea level at 2850m. The inverted index matched the "Featherleaf Clove" vegetation tag to select entries related to this habitat. The full-text index matched the "territory declaration" behavior tag to exclude irrelevant entries (such as migratory behavior explanations). A temperature of 18°C ​​was mapped to the database range "10-25°C." Location coordinates were geocoded to the "Featherleaf Clove distribution area" (predefined based on the protected area vegetation GIS layer). The behavior "territory declaration" was converted to the database tag "breeding season territory marker." The environmental B-tree index combined with the inverted index identified 15 candidate entries with "2000-3500m above sea level and vegetation containing Featherleaf Clove." The full-text index further filtered out entries containing the "territory declaration" and "high-altitude gliding" tags, leaving 8 entries. Finally, a cosine similarity calculation (comparing the matching degree of the tag factor with the environmental conditions of the entry) identified highly relevant explanation entries.

[0078] Item 1: Species classification of golden eagles

[0079] Trigger Condition: Species = Golden Eagle

[0080] Narration: The golden eagle belongs to the class Aves, order Accipitridae, family Accipitridae, and genus Accipitridae. It is a large bird of prey that occupies a crucial position in avian evolution. From a taxonomic perspective, its unique physical characteristics and behavioral habits distinguish it from other species in the same genus. Its powerful wings and sharp talons are key features of its adaptation to predatory behavior, making it a typical example of the Accipitridae family.

[0081] Entry 2: Habitat of the Golden Eagle

[0082] Trigger conditions: Species = Golden Eagle, Vegetation = Syringa syringae, Altitude 2000-3500m

[0083] Narration: In Gansu's Liancheng Nature Reserve, golden eagles favor environments like the current Syringa syringa distribution area. Syringa syringa thickets provide excellent habitat and breeding conditions for golden eagles. These 2-3 meter tall thickets, with a moderate density of branches and leaves, provide shelter for their nests while allowing them to observe their surroundings and search for prey from their perch. Furthermore, at an altitude of 2,000-3,500 meters, the relatively thin air and wide field of vision in this area are ideal for golden eagles' high-altitude gliding and hunting activities. Research has shown that approximately 70% of golden eagle nests in the reserve are within 50 meters of Syringa syringa thickets, highlighting the eagles' deep dependence on this vegetation.

[0084] Item 3: Endangered status of golden eagles

[0085] Trigger condition: Species = Golden Eagle

[0086] Narration content: ...

[0087] Entry 4: Legends of the Golden Eagle

[0088] Trigger condition: Species = Golden Eagle

[0089] Narration content: ...

[0090] Item 5: Ecological value of golden eagles

[0091] Trigger conditions: Species = Golden Eagle, Behavior = Predation

[0092] Narration: As a top predator, the golden eagle plays a vital role in the ecosystem. It primarily feeds on small and medium-sized mammals such as hares and blue sheep. By preying on these prey, it controls their populations and maintains the ecosystem's balance. For example, in the range of Syringa syringae, the golden eagle's predation keeps hare populations within a reasonable range (approximately 5 individuals per hectare), preventing overpopulation and damage to local herbaceous plants (such as Stipa), thereby indirectly protecting the regeneration capacity of vegetation. Furthermore, the presence of the golden eagle influences the behavior and distribution of other organisms, positively contributing to the stability and diversity of the entire ecosystem.

[0093] Item 6: Evolutionary history of the golden eagle

[0094] Trigger condition: Species = Golden Eagle

[0095] Narration: The golden eagle's evolutionary history dates back millions of years. During this long evolutionary process, it gradually adapted to a variety of complex environments. Early ancestors of eagles were likely small, with limited flight capabilities and hunting techniques. Over time, to better survive in a highly competitive ecosystem, the golden eagle's ancestors gradually evolved larger bodies, stronger wings, and sharper claws.

[0096] Finally, the above explanation items are input into the large model to generate the explanation content: ...

[0097] The integrity and accuracy of the interpretation database are crucial. Failure to update the database promptly, or errors or outdated information in the interpretation content, will compromise the quality of the final interpretation. Furthermore, for some complex ecological phenomena or newly emerging species and their environmental relationships, the database may lack corresponding interpretation entries, resulting in an inability to provide comprehensive interpretations. By establishing clear indexes or association rules, we can quickly match relevant interpretation content.

[0098] The fuzzy logic algorithm is specifically as follows: several fuzzy sets are defined for each environmental factor, and the membership function is determined based on each fuzzy set. The membership function is used to describe the degree to which the marking factor belongs to the corresponding fuzzy set. The membership of the marking factor in the corresponding fuzzy set is calculated, and the combined membership is obtained according to the fuzzy logic operation rules. If the combined membership is greater than the pre-set trigger threshold, the corresponding interpretation item is triggered.

[0099] In view of the fuzziness of environmental parameters (such as "high temperature" and "dryness" have no absolute boundaries), continuous parameters are mapped to membership degrees (0-1) through fuzzy sets. The following example shows:

[0100] 1. For the temperature factor "high temperature", define the triangular membership function:

[0101]

[0102] When the temperature is ≥30℃, the membership is 1 (absolute high temperature), the membership increases linearly between 25-30℃, and the membership is 0 when ≤25℃.

[0103] Humidity factor "Dry"

[0104]

[0105] When humidity is ≤30%, the membership is 1 (absolute dryness), it decreases linearly between 30% and 40%, and it is 0 when it is ≥40%.

[0106] When multiple environmental factors need to be combined (such as "temperature > 30°C and humidity < 40%"), boundary fuzzy processing is achieved through fuzzy logic operation:

[0107] Substitute the real-time data (such as current T = 28 ° C, RH = 35%) into the membership function and calculate the single factor membership: μ high temperature (28) = 0.6, μ dry (35) = 0.5

[0108] Activate the association rule (e.g., "high temperature ∧ dryness → trigger the entry 'high temperature adaptability of desert ecosystems'") and use the minimum operator to calculate the combined membership: μtrigger = min(μhigh temperature, μdryness) = 0.5; set a trigger threshold (e.g., ≥0.4). When the combined membership reaches the standard, activate the corresponding interpretation entry.

[0109] Explanation entry template: The current temperature is {temp}℃, the humidity is {rh}%, and the Haloxylon ammodendron adapts to the environment through {adaptation}, and its transpiration rate increases by {rate}% compared with the humid area.

[0110] The explanation entry is set with an explanation template including parameter placeholders. The tag factors are filled into the parameter placeholders and the content of the explanation template is updated. The predefined structured template contains dynamic parameters and supports three placeholders: value, text, and formula:

[0111] Numeric placeholders: such as {temp} (temperature value), {wind_speed} (wind speed);

[0112] Text placeholder: such as {adaptation_mechanism} (adaptation mechanism name);

[0113] Formula placeholder: such as {transpiration_rate=0.8*wind_speed+5} (transpiration rate calculation formula).

[0114] Explanation template: The current wind speed is {wind_speed} m / s, which falls under {wind_level} (wind speed classification: <3 m / s = breeze, 3-8 m / s = moderate wind, >8 m / s = strong wind). In strong winds, the leaf pitch of a Populus euphratica tree automatically adjusts to {angle}° to reduce evaporation, saving {save_water}% water compared to calm winds. Real-time parameters are acquired through sensors (e.g., a wind speed sensor outputs 5 m / s). After preprocessing (e.g., denoising and unit conversion), the basic parameters are directly inserted into numerical placeholders (e.g., wind_speed = 5). Derived parameters are calculated using preset formulas (e.g., wind_level = moderate wind, angle = 30 - 0.5 * wind_speed = 27.5°, save_water = 10 + 2 * wind_speed = 20%). Text placeholders are dynamically generated based on species characteristics (e.g., "Populus euphratica" corresponds to "deep root system for water absorption," "Haloxylon ammodendron" corresponds to "degenerate leaves into scales"). The output is: "The current wind speed is 5m / s, which is considered a moderate wind. In strong winds, the leaf angle of the poplar tree automatically adjusts to 27.5° to reduce water evaporation, saving 20% ​​of water compared to calm winds."

[0115] It also includes that a monitoring point is set up in the target area. When the monitoring point detects a trigger signal, the broadcast of the corresponding commentary content is triggered. The trigger signal is set as a human approach signal. The human approach signal is obtained through an infrared sensor, Bluetooth or WIFI probe. If the human body approaches at a distance of less than 10 meters, a trigger instruction is generated and transmitted to the commentary system through a wired / wireless communication module to activate the broadcast of the corresponding commentary content.

[0116] The present invention collects images, sounds and environmental data through devices such as infrared cameras and microphone arrays, and combines convolutional neural networks (CNNs), bidirectional LSTM networks and Bayesian voting algorithms to achieve multimodal data complementarity, thereby improving the accuracy and robustness of species identification. By constructing a species-environmental factor knowledge graph, the present invention uses Spearman correlation analysis, canonical correspondence analysis (CCA) and fuzzy logic algorithms to explore the deep relationship between species and the environment, greatly improving the scientific nature of the determination of interpretation objects. Fuzzy logic is used to handle the fuzzy boundary problems of environmental parameters, and the trigger conditions are more in line with the real ecological scene. The matching accuracy of interpretation items is improved compared with the traditional threshold method. By associating the interpretation database with marker factors, combining large models and dynamic parameter injection, personalized interpretations containing real-time environmental parameters are generated, and dynamic data is linked to generate personalized interpretation content. The monitoring point trigger mechanism solves the interpretation lag problem of traditional guide systems.

[0117] Example 2

[0118] Real-time environmental data linkage interpretation control system, including:

[0119] Data collection units, including infrared cameras, drone cameras, or fixed surveillance cameras for collecting image data, microphone arrays, bird song identifiers, or underwater sonar for collecting sound data, and weather stations, geographic information systems, and air quality monitors for obtaining abiotic environmental data;

[0120] The species identification unit uses a deep learning model trained with species samples to receive data from the data acquisition unit, identify and determine species information, and output the results to the environmental analysis unit;

[0121] The environmental analysis unit is used to extract environmental factors, call the knowledge graph to perform correlation analysis between species information and environmental factors, and determine the interpretation object and transmit it to the interpretation association unit;

[0122] An interpretation association unit is used to record the field data corresponding to the interpretation object as a marking factor, associate it with the interpretation database to obtain an interpretation entry, and send the interpretation entry to the content generation unit; and

[0123] The content generation unit is used to input interpretation items into the large model to generate interpretation content, optimize it in combination with field data and historical data, and output the generated content.

[0124] Example 3

[0125] It also includes an update feedback unit, which is used to update the deep learning model, knowledge graph and interpretation database, and feed back to the content generation unit to make real-time adjustments to the generated content.

[0126] In a preferred embodiment, the present invention further discloses an electronic device, comprising:

[0127] a memory for storing computer programs executable on the processor;

[0128] The processor is used to execute the computer program stored in the memory to implement the above real-time environmental data linkage interpretation control method.

[0129] like Figure 2 Components of the electronic device may include, but are not limited to, one or more processors 410, a memory 430, and a communication bus 440 connecting different system components (including the memory 430 and the processing unit 410).

[0130] Communication bus 440 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnection (PCI) bus.

[0131] Electronic devices typically include a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device, including volatile and non-volatile media, removable and non-removable media.

[0132] Memory 430 may include computer-readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The electronic device may further include other removable / non-removable, volatile / non-volatile computer system storage media. Memory 430 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of various embodiments of the present invention.

[0133] A program / utility having a set (at least one) of program modules may be stored in memory 430. Such program modules include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The program modules generally perform the functions and / or methods described in the embodiments of the present invention.

[0134] The processor 410 executes various functional applications and data processing by running the programs stored in the memory 430, such as implementing the embodiments of the present invention. Figure 2 The illustrated embodiment provides a method for managing prefabricated decoration information.

[0135] In a preferred embodiment, the present invention further discloses a processor executable program, which can be directly executed by a processor or can be executed by a processor after being compiled, and is used to execute the real-time environmental data linkage interpretation control method as described above.

[0136] In a preferred embodiment, the present invention further discloses a storage medium, characterized in that the storage medium stores non-volatile program code that is executable by a processor or executable after compilation, and the program code is used to execute the real-time environmental data linkage interpretation control method as described above.

[0137] The above-mentioned computer-readable storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM) or flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or device.

[0138] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0139] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0140] The computer program code for performing the operation of the embodiments of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (Local Area Network; hereinafter referred to as: LAN) or a wide area network (Wide Area Network; hereinafter referred to as: WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0141] The foregoing description describes specific embodiments of the present invention. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0142] In the description of the embodiments of the present invention, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" 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 embodiments of the present invention. In the embodiments of the present invention, 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 different embodiments or examples described in the embodiments of the present invention and the features of different embodiments or examples, unless they are mutually inconsistent.

[0143] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. In the description of the embodiments of the present invention, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0144] 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 implementation of the embodiments of the invention includes alternative implementations in which functions may be performed out of 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 those skilled in the art to which the embodiments of the invention pertain.

[0145] The word "if," as used herein, may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0146] It should be noted that the terminals involved in the embodiments of the present invention may include but are not limited to personal computers (Personal Computer; hereinafter referred to as: PC), personal digital assistants (Personal Digital Assistant; hereinafter referred to as: PDA), wireless handheld devices, tablet computers (Tablet Computer), mobile phones, MP3 players, MP4 players, etc.

[0147] In the several embodiments provided in the embodiments of the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection of the devices or units through some interfaces, which may be electrical, mechanical or other forms.

[0148] In addition, the functional units in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional units.

[0149] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to perform some steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, an optical disk, and other media that can store program code.

[0150] The above description is only a preferred embodiment of the embodiment of the present invention and is not intended to limit the embodiment of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the embodiment of the present invention should be included in the scope of protection of the embodiment of the present invention.

Claims

1. A real-time environmental data linkage commentary control method, characterized in that: include: Field data acquired in the target area is combined with deep learning models to determine the corresponding species information; Extract environmental factors, call the knowledge graph to perform correlation analysis between the species information and environmental factors, and determine the interpretation object; Recording the on-site data corresponding to the interpretation object as a marking factor, and associating the marking factor with the interpretation database to obtain an interpretation entry; Input the explanation entries into the large model to generate the explanation content.

2. The real-time environmental data linkage commentary control method according to claim 1, characterized in that: The field data includes image data, sound data and non-biological environmental data. The image data is collected by an infrared camera, drone camera or fixed monitoring camera arranged in the target area. The sound data is collected by a microphone array, bird song identifier or underwater sonar. The non-biological environmental data is obtained in real time through a weather station, geographic information system and air quality monitor.

3. The real-time environmental data linkage commentary control method according to claim 1, characterized in that: Species information includes the species name and the species quantity, morphological characteristics, species behavior and location information associated with the species name. The species quantity is determined by matching the target detection model or tracking algorithm, the morphological characteristics are determined by feature matching through ResNet combined with the knowledge graph, the species behavior is determined by the time series model combined with the knowledge graph mapping behavior label, and the location information is determined by GPS positioning.

4. The real-time environmental data linkage commentary control method according to claim 2, characterized in that: The species information is specifically determined by the following steps: pre-processing the image data and the sound data, extracting image features using a convolutional neural network model, and extracting sound features using a bidirectional LSTM network; Bayesian voting is performed on the output results of the convolutional neural network and the bidirectional LSTM network, and the fusion rules are set as follows: if the confidence in the convolutional neural network and the bidirectional LSTM network is greater than the species recognition threshold and the species name is consistent, the species name is output; the confidence of the convolutional neural network and the bidirectional LSTM network is sorted and the joint probability is determined, and the species name corresponding to the maximum joint probability is output; if the joint probabilities are all less than the joint lower limit threshold, a no data result is output.

5. The real-time environmental data linkage commentary control method according to claim 4, characterized in that: The output results of the convolutional neural network and the bidirectional LSTM network are verified for spatiotemporal consistency. The azimuth of the output sound source is estimated through the time delay of the microphone array and cross-validated with the position information of the image data to detect whether the image data and the sound data are both within the target area.

6. The real-time environmental data linkage commentary control method according to claim 5, characterized in that: Environmental factors are obtained from field data within the target area, and the environmental factors include altitude, slope, aspect and intensity of human disturbance. The knowledge graph is called to perform correlation analysis between the species information and the environmental factors, specifically: constructing a species-environmental factor matrix, determining the correlation between species information and environmental factors through Spearman correlation analysis and / or canonical correspondence analysis, and establishing a species distribution prediction model based on the correlation. The species distribution prediction model is verified by a Monte Carlo permutation test, and the total explanatory capacity of environmental factors is greater than 30%.

7. The real-time environmental data linkage commentary control method according to claim 6, characterized in that: The process of associating the marking factor with the interpretation database to obtain the interpretation entries is as follows: first, a hash index algorithm is used to screen the interpretation entries of a single marking factor, and then a fuzzy logic algorithm is used to process the interpretation entries corresponding to multiple marking factors.

8. The real-time environmental data linkage commentary control method according to claim 7, characterized in that: The fuzzy logic algorithm includes: defining several fuzzy sets for each of the environmental factors, determining a membership function based on each of the fuzzy sets, wherein the membership function is used to describe the degree to which the marking factor belongs to the corresponding fuzzy set, calculating the membership of the marking factor in the corresponding fuzzy set, and obtaining the combined membership according to the fuzzy logic operation rules. If the combined membership is greater than a preset trigger threshold, the corresponding interpretation item is triggered.

9. The real-time environmental data linkage commentary control method according to claim 8, characterized in that: The explanation entry is provided with an explanation template including a parameter placeholder, the marking factor is filled into the parameter placeholder, and the content of the explanation template is updated.

10. The real-time environmental data linkage commentary control method according to claim 9, characterized in that: It also includes that the target area is provided with a monitoring point, and when the monitoring point detects a trigger signal, the broadcast of the corresponding commentary content is triggered, and the trigger signal is set to a human approach signal, and the human approach signal is obtained through an infrared sensor, Bluetooth or WIFI probe.

11. Real-time environmental data linkage interpretation control system, characterized by: include: Data collection units, including infrared cameras, drone cameras, or fixed surveillance cameras for collecting image data, microphone arrays, bird song identifiers, or underwater sonar for collecting sound data, and weather stations, geographic information systems, and air quality monitors for obtaining abiotic environmental data; a species identification unit, which receives data from the data acquisition unit using a deep learning model trained with species samples, identifies and determines species information, and outputs the results to the environmental analysis unit; The environmental analysis unit is used to extract environmental factors, call the knowledge graph to perform correlation analysis on the species information and environmental factors, determine the interpretation object and transmit it to the interpretation association unit; An interpretation association unit is used to record the field data corresponding to the interpretation object as a marking factor, associate it with the interpretation database to obtain an interpretation entry, and send the interpretation entry to the content generation unit; as well as The content generation unit is used to input interpretation items into the large model to generate interpretation content, optimize it in combination with field data and historical data, and output the generated content.

12. The real-time environmental data linked interpretation control system according to claim 11, characterized in that: It also includes an update feedback unit for updating the deep learning model, knowledge graph and interpretation database, and feeding back to the content generation unit to make real-time adjustments to the generated content.

13. An electronic device, characterized in that: include: a memory for storing computer programs executable on the processor; A processor is used to execute the computer program stored in the memory to implement the real-time environmental data linkage interpretation control method according to any one of claims 1 to 10.

14. A processor executable program, wherein the processor executable program can be directly executed by a processor or can be executed by a processor after being compiled, and is used to execute the real-time environmental data linkage interpretation control method according to any one of claims 1 to 10.

15. A storage medium, characterized in that: The storage medium stores non-volatile program code that is executable by a processor or executable after compilation, and the program code is used to execute the real-time environmental data linkage interpretation control method according to any one of claims 1 to 10.

Images