Monitoring method and system based on artificial intelligence

Abstract

The invention discloses a monitoring method and system based on artificial intelligence, and relates to the technical field of artificial intelligence, and the monitoring method based on artificial intelligence comprises the following steps: S1, obtaining raw material real-time data according to a preset collection frequency; s2, analyzing the real-time data of the raw materials to obtain raw material analysis data; s3, analyzing the raw material analysis data through a preset analysis condition to generate a monitoring result, and executing S4 when the monitoring result is to-be-supplemented; when the monitoring result is that supplement is not needed, S1 is executed; and S4, automatically generating a raw material purchasing order, sending the raw material purchasing order to the enterprise terminal, and after confirmation information sent by the enterprise terminal according to the raw material purchasing order is received, broadcasting an encryption request to the supplier node to complete the raw material purchasing order. According to the invention, the accuracy and real-time performance of the monitoring result can be improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and particularly to a monitoring method and system based on artificial intelligence. Background Art

[0002] In the field of industrial production, raw material inventory management is a core link to ensure the continuity of production. The existing technologies mainly rely on the following methods to achieve raw material monitoring and replenishment:

[0003] 1. Single-sensor monitoring: Using a weight sensor or manual inspection to record the inventory level to achieve raw material monitoring and replenishment. However, this method has problems such as being unable to identify abnormal physical forms of raw materials (such as caking, leakage), and errors are easily introduced in manual operations. For example, an enterprise may cause the production line to stop due to the failure to detect raw material caking in a timely manner.

[0004] 2. Static threshold warning: Using a replenishment reminder triggered based on a fixed safety threshold to achieve raw material monitoring. However, this method does not consider dynamic factors such as equipment status and environmental parameters. Statistics show that the false alarm rate of this method is as high as 35%, and the risk of raw material shortage in future periods cannot be predicted.

[0005] 3. Discretized data processing: Independently processing the raw material data collected by sensors, image acquisition devices, or RFID tag readers to obtain raw material monitoring results. This method lacks a multi-source information fusion mechanism and cannot correlate production plans with raw material consumption data, which will lead to a mismatch between the procurement cycle and production demand and a decline in inventory turnover.

[0006] 4. Dependence on manual decision-making: All links such as the generation of raw material purchase orders and supplier verification rely on manual operations, with a long response time and a risk of human error. For example, it is easy to cause the raw material specifications to not match.

[0007] Therefore, there is an urgent need to propose a brand-new monitoring method and system based on artificial intelligence to solve the above problems. Summary of the Invention

[0008] The purpose of this application is to provide a monitoring method and system based on artificial intelligence, which can improve the accuracy and real-time performance of monitoring results.

[0009] To achieve the above object, the present application provides an artificial intelligence-based monitoring method, including the following steps: S1: Obtain real-time raw material data according to a preset acquisition frequency. Among them, the real-time raw material data at least includes: current raw material inventory data, raw material physical form change data, and raw material batch information; S2: Analyze the real-time raw material data to obtain raw material analysis data. Among them, the raw material analysis data at least includes: current inventory, predicted inventory data, and abnormal loss data; S3: Analyze the raw material analysis data through preset analysis conditions to generate a monitoring result, where the monitoring result is to be replenished or not to be replenished; when the monitoring result is to be replenished, execute S4; when the monitoring result is not to be replenished, execute S1; S4: Automatically generate a raw material purchase order and send the raw material purchase order to the enterprise side. After receiving the confirmation information sent by the enterprise side according to the raw material purchase order, broadcast an encryption request to the supplier node to complete the raw material purchase order.

[0010] As above, among them, the sub-steps of analyzing the real-time raw material data to obtain the raw material analysis data are as follows: S21: Preprocess the real-time raw material data to obtain preprocessed data. Among them, the preprocessed data includes: filtered current inventory, raw material form abnormal data, and corrected batch information; S22: Use the filtered current inventory in the preprocessed data as the current inventory; S23: Construct input variables according to the preprocessed data, temperature parameters, and humidity parameters, and input the input variables into a pre-constructed prediction model. The prediction model outputs the predicted inventory for the next N time instants as the predicted inventory data; S24: Analyze the preprocessed data to obtain abnormal loss data. Among them, the abnormal loss data at least includes: raw material feature similarity, caking area ratio, leakage probability, and leakage duration; S25: Use the current inventory, predicted inventory data, and abnormal loss data as the raw material analysis data.

[0011] As above, among them, the sub-steps of preprocessing the real-time raw material data to obtain the preprocessed data are as follows: S211: Filter the current raw material inventory data in the real-time raw material data through the Kalman filter algorithm to obtain the filtered current inventory; S212: Extract features from the raw material physical form change data in the real-time raw material data through the target detection algorithm to obtain raw material form abnormal data. Among them, the raw material form abnormal data at least includes: caking index, leakage probability, looseness, and stacking stability index; S213: Correct the transmission error of the raw material batch information in the real-time raw material data to obtain the corrected batch information; S214: Use the filtered current inventory, raw material form abnormal data, and corrected batch information as the preprocessed data.

[0012] As above, among them, the expression of the input variable is: Among them, is the input variable, tc Indicates the time point for collecting real-time data of raw materials; Qdqkc is the filtered current inventory in the preprocessed data; Qwlzs is the physical form conversion coefficient; Pllxh is the theoretical consumption in the next N moments in the production plan scheduling, where N is a natural number; Ewd is the temperature parameter of the raw material storage area; Esd is the humidity parameter of the raw material storage area.

[0013] As described above, among them, the sub-steps of analyzing the raw material analysis data through preset analysis conditions to generate monitoring results are as follows: S31: Use the preset inventory safety threshold to judge the current inventory and generate a monitoring result. If the current inventory is less than or equal to the preset inventory safety threshold, the generated monitoring result is to be replenished; if the current inventory is greater than the preset inventory safety threshold, execute S32; S32: Use the preset abnormal damage conditions to judge the abnormal loss data and generate a monitoring result. If the raw material feature similarity in the abnormal loss data is less than the raw material feature similarity threshold in the abnormal damage conditions, and the proportion of the caking area in the abnormal loss data is greater than the caking area proportion threshold in the abnormal damage conditions, the generated monitoring result is to be replenished, and / or if the leakage probability in the abnormal loss data is greater than the leakage probability threshold in the abnormal damage conditions, and the leakage duration in the abnormal loss data is greater than the leakage duration threshold in the abnormal damage conditions, the generated monitoring result is to be replenished; if the raw material feature similarity in the abnormal loss data is greater than or equal to the raw material feature similarity threshold in the abnormal damage conditions, the proportion of the caking area in the abnormal loss data is less than the caking area proportion threshold in the abnormal damage conditions, the leakage probability in the abnormal loss data is less than or equal to the leakage probability threshold in the abnormal damage conditions, and / or the leakage duration in the abnormal loss data is less than the leakage duration threshold in the abnormal damage conditions, execute S33; S33: Analyze the predicted inventory data using the dynamically adjusted threshold and generate a monitoring result. If there is a predicted inventory in the predicted inventory data that is less than the dynamically adjusted threshold, the generated monitoring result is to be replenished; if each predicted inventory in the predicted inventory data is greater than or equal to the dynamically adjusted threshold, the generated monitoring result is no need to replenish.

[0014] As described above, among them, the expression of the dynamically adjusted threshold is: Among them, is the nth predicted inventory in the predicted inventory data The corresponding dynamically adjusted threshold; T s is the fixed safety threshold; β is the form change influence factor; EMA(·) is the exponential smoothing operator; Qwlzs is the physical form conversion coefficient; α sb is the industrial production equipment status coefficient; σ ls is the standard deviation of the raw material consumption in the past f days, where f is a natural number; μ lsis the average consumption of raw materials in the past f days.

[0015] As described above, wherein, a raw material purchase order is automatically generated through a blockchain smart contract, and the raw material purchase order is sent to the enterprise side. After receiving the confirmation information sent by the enterprise side according to the raw material purchase order, an encrypted request is broadcast to the supplier node to complete the raw material purchase order.

[0016] This application also provides an artificial intelligence-based monitoring system, including: a data acquisition subsystem, an artificial intelligence monitoring center, and an enterprise side; wherein, the data acquisition subsystem: is used to collect real-time raw material data according to a preset acquisition frequency and send the real-time raw material data to the artificial intelligence monitoring center; the artificial intelligence monitoring center: is used to execute the above-mentioned artificial intelligence-based monitoring method; the enterprise side: is used to receive the raw material purchase order and send a confirmation information according to the raw material purchase order.

[0017] As described above, wherein, the data acquisition subsystem at least includes: a weight sensor, an environmental sensor, an image acquisition device, and an RFID tag reader; wherein, the weight sensor: is used to collect the current inventory data of raw materials; the environmental sensor: is used to collect the temperature parameter of the raw material storage area and the humidity parameter of the raw material storage area; the image acquisition device: is used to obtain the physical form change data of raw materials; the RFID tag reader: is used to obtain the raw material batch information.

[0018] As described above, wherein, the image acquisition device is an industrial camera.

[0019] The beneficial effects achieved by this application are as follows:

[0020] (1) The artificial intelligence-based monitoring method and system of this application can improve the accuracy and real-time performance of monitoring results

[0021] (2) The artificial intelligence-based monitoring method and system of this application realize real-time monitoring and intelligent replenishment of industrial raw materials through multi-source data fusion, and can solve the problems of monitoring blind spots and insufficient accuracy caused by a single data source.

[0022] (3) The artificial intelligence-based monitoring method and system of this application solve the adaptability contradiction between the static threshold warning mechanism and the dynamic environmental factors by dynamically adjusting the threshold.

[0023] (4) The artificial intelligence-based monitoring method and system of this application can avoid the supply chain response bottleneck caused by the lag of manual decision-making and the security risks in cross-system data interaction. Description of the Drawings

[0024] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0025] Figure 1 It is a schematic structural diagram of an embodiment of a monitoring system based on artificial intelligence;

[0026] Figure 2 It is a flowchart of an embodiment of a monitoring method based on artificial intelligence. Detailed implementation manners

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0028] As Figure 1 shown, the present application provides a monitoring system based on artificial intelligence, including: a data acquisition subsystem 1, an artificial intelligence monitoring center 2, and an enterprise end 3.

[0029] Among them, the data acquisition subsystem 1: is used to collect real-time raw material data according to a preset acquisition frequency and send the real-time raw material data to the artificial intelligence monitoring center 2.

[0030] The artificial intelligence monitoring center 2: is used to execute the following monitoring method based on artificial intelligence.

[0031] The enterprise end 3: is used to receive raw material purchase orders and send confirmation information according to the raw material purchase orders.

[0032] Furthermore, the data acquisition subsystem 1 at least includes: a weight sensor, an environmental sensor, an image acquisition device, and an RFID tag reader.

[0033] Among them, the weight sensor: is used to collect the current inventory data of raw materials.

[0034] The environmental sensor: is used to collect the temperature parameter of the raw material storage area and the humidity parameter of the raw material storage area.

[0035] The image acquisition device: is used to obtain the data of the physical form change of raw materials.

[0036] The RFID tag reader: is used to obtain the batch information of raw materials.

[0037] Further, the image acquisition device is an industrial camera, but is not limited to an industrial camera. Preferably, the present application uses an industrial camera.

[0038] As Figure 2 shown, the present application provides a monitoring method based on artificial intelligence, including the following steps:

[0039] S1: Obtain real-time raw material data according to a preset acquisition frequency. Among them, the real-time raw material data at least includes: current inventory data of raw materials, physical form change data of raw materials, and batch information of raw materials.

[0040] Specifically, the specific value of the preset acquisition frequency is set according to the actual situation.

[0041] Since the raw material characteristics and uses of different types of raw materials are different, preferably: the preset acquisition frequencies for different types of raw materials are different, which can improve the real-time monitoring of the corresponding raw materials.

[0042] The current inventory data of raw materials is the quantity of available raw materials actually held by the enterprise at the current acquisition time node.

[0043] The physical form change data of raw materials is the core index reflecting the dynamic changes of the material properties of raw materials, including: state parameters, geometric parameters, mechanical parameters, thermal parameters, optical parameters, and electrical parameters, etc. State parameters include phase transition, dispersion state, and surface state. Geometric parameters include size change, shape characteristics, volume / density. Mechanical parameters include mechanical properties, rheological characteristics, and friction coefficient. Thermal parameters include temperature field distribution and thermal expansibility. Optical parameters include color change and glossiness. Electrical parameters include conductivity and magnetic characteristics.

[0044] The batch information of raw materials is the core data unit for realizing full-process traceability, quality control, and production optimization, including: unique coding (such as batch number, traceability code, and supplier internal batch number), source information (such as supplier code, purchase order number, and place of origin), production data (such as production date, production team number, and production equipment number), quality verification (such as inspection report number, key index value, and inspection agency), warehousing data (such as warehousing time, warehouse location number, and storage conditions), transportation records (such as logistics company name, transportation order number, and transportation environment data), production association (such as assigned production order number, used work station number, and feeding time), and consumption statistics (such as total usage, remaining inventory, and consumption speed per unit time), etc.

[0045] S2: Analyze the real-time raw material data to obtain raw material analysis data. Among them, the raw material analysis data at least includes: current inventory, predicted inventory data, and abnormal loss data.

[0046] Further, the sub-steps for analyzing the real-time raw material data to obtain the analyzed raw material data are as follows:

[0047] S21: Preprocess the real-time raw material data to obtain the preprocessed data. The preprocessed data includes: the current inventory quantity after filtering, the abnormal raw material form data, and the corrected batch information.

[0048] Further, the sub-steps for preprocessing the real-time raw material data to obtain the preprocessed data are as follows:

[0049] S211: Filter the current inventory quantity data of the raw materials in the real-time raw material data through the Kalman filtering algorithm to obtain the current inventory quantity after filtering.

[0050] Specifically, the Kalman filtering algorithm can eliminate the time-series noise in the current inventory quantity data of the raw materials, but is not limited to the Kalman filtering algorithm.

[0051] Further, filter the current inventory quantity data of the raw materials in the real-time raw material data through the Kalman filtering algorithm of the second-order adaptive model to obtain the current inventory quantity after filtering. Among them, the initial values of the process noise matrix Q and the measurement noise matrix R of the Kalman filtering algorithm of the second-order adaptive model satisfy:

[0052]

[0053] where σ w is the historical noise standard deviation of the weight sensor for collecting the current inventory quantity data of the raw materials.

[0054] S212: Extract the features of the physical form change data of the raw materials in the real-time raw material data through the object detection algorithm to obtain the abnormal raw material form data. The abnormal raw material form data includes at least: the caking index, the leakage probability, the looseness, and the stacking stability index.

[0055] Further, the YOLOv5 model is used for the object detection algorithm, but is not limited to the YOLOv5 model.

[0056] S213: Correct the transmission error of the batch information of the raw materials in the real-time raw material data to obtain the corrected batch information.

[0057] Specifically, the cyclic redundancy check (CRC) is used to correct the transmission error of the batch information of the raw materials in the real-time raw material data to obtain the corrected batch information, but is not limited to the cyclic redundancy check (CRC). The cyclic redundancy check (CRC) is preferably used in this application to ensure data integrity.

[0058] S214: Use the current inventory quantity after filtering, the abnormal raw material form data, and the corrected batch information as the preprocessed data.

[0059] S22: Use the filtered current inventory in the preprocessed data as the current inventory.

[0060] S23: Construct input variables based on the preprocessed data, temperature parameters, and humidity parameters, and input the input variables into a pre-constructed prediction model. The prediction model outputs the predicted inventory for the next N time points as the predicted inventory data.

[0061] Furthermore, the expression for the input variable is:

[0062]

[0063] where is the input variable, and t c represents the time point for collecting real-time raw material data; Qdqkc is the filtered current inventory in the preprocessed data; Qwlzs is the physical form conversion coefficient; Pllxh is the theoretical consumption for the next N time points in the production plan scheduling, where N is a natural number; Ewd is the temperature parameter of the raw material storage area; Esd is the humidity parameter of the raw material storage area.

[0064] Specifically, the specific value of N is set according to the actual situation. In this application, preferably: N = 8. When t c , Ewd and Esd are collected by environmental sensors installed in the raw material storage area.

[0065] Furthermore, the expression for the physical form conversion coefficient Qwlzs is:

[0066] Qwlzs = η·Zjk + ζ·Pxl + λ·Dss + δ·EMA(ΔS);

[0067] where η is the caking influence factor; Zjk is the caking index in the raw material form abnormal data; ζ is the leakage correction coefficient; Pxl is the leakage probability in the raw material form abnormal data; λ is the loose compensation rate; Dss is the looseness in the raw material form abnormal data; δ is the stability decay factor; EMA(·) is the exponential smoothing operator; ΔS is the stacking stability index in the raw material form abnormal data.

[0068] Specifically, η·Zjk represents the caking effect; ζ·Pxl represents the leakage loss; λ·Dss represents the loose compensation; δ·EMA(ΔS) represents the stacking stability. The individual Pxl, ΔS, and Dss can be obtained using existing calculation formulas, so they will not be elaborated here.

[0069] Furthermore, the expression for the caking index Zjk in the raw material form abnormal data is:

[0070]

[0071] Specifically, Ajk i is the area of the i-th agglomerate region obtained by feature extraction of the raw material physical form change data in the real-time raw material data by the target detection algorithm, where i ∈ [1, I], I is the total number of agglomerate regions, and I is a natural number; Azg is the area of all regions covered by the raw material in the storage area; Qxd i is the sphericity of the i-th agglomerate region; Esd is the humidity parameter of the raw material storage area; e is the base of the natural logarithm.

[0072] Furthermore, the sphericity Qxd of the i-th agglomerate region is calculated by Hough circle detection i , but not limited to Hough circle detection.

[0073] Furthermore, the YOLOv5 model is used to extract features from the raw material physical form change data in the real-time raw material data, and the area Ajk of the i-th agglomerate region is obtained by segmentation i .

[0074] Specifically, the EMA is introduced into the expression of the physical form conversion coefficient to smooth the rapidly changing stacking stability, which complements the real-time detection of the YOLOv5 model, improves the time resolution in the raw material monitoring process, and improves the prediction accuracy of the predicted inventory data.

[0075] Furthermore, the expression of the n-th predicted inventory in the predicted inventory data is: where n ∈ [1, N], and t c represents the time point when the real-time raw material data is collected.

[0076] Furthermore, the prediction model uses a pre-trained LSTM neural network, but not limited to a pre-trained LSTM neural network. This application preferably uses a pre-trained LSTM neural network.

[0077] Specifically, the training of the LSTM neural network is completed by a training data set composed of historical raw material consumption records, production plan scheduling data, and correlation parameters collected by environmental temperature and humidity sensors. The existing training methods can be used to achieve this, so it will not be elaborated here.

[0078] S24: Analyze the preprocessed data to obtain abnormal loss data, where the abnormal loss data at least includes: raw material feature similarity, agglomerate region ratio, leakage probability, and leakage duration.

[0079] Specifically, the raw material feature similarity is the similarity between the raw material feature data extracted from the preprocessed data and the raw material feature data under normal conditions.

[0080] The proportion of the caking area is the ratio between the total area of the caking area of the raw materials obtained from the pre-processed data and the area of the entire area covered by the raw materials, that is: the total area of the caking area of the raw materials / the area of the entire area covered by the raw materials.

[0081] The leakage probability is the leakage probability in the abnormal raw material form data.

[0082] The leakage duration is the duration between the moment when the raw materials leak and the time point when the real-time data of the raw materials is collected.

[0083] S25: Use the current inventory, predicted inventory data, and abnormal loss data as raw material analysis data.

[0084] S3: Analyze the raw material analysis data through preset analysis conditions to generate a monitoring result, where the monitoring result is to be replenished or not to be replenished; when the monitoring result is to be replenished, execute S4; when the monitoring result is not to be replenished, execute S1.

[0085] Furthermore, the sub-steps of analyzing the raw material analysis data through preset analysis conditions to generate a monitoring result are as follows:

[0086] S31: Use the preset inventory safety threshold to judge the current inventory and generate a monitoring result. If the current inventory is less than or equal to the preset inventory safety threshold, the generated monitoring result is to be replenished; if the current inventory is greater than the preset inventory safety threshold, execute S32.

[0087] Specifically, the specific value of the preset inventory safety threshold is set according to the actual situation of the enterprise.

[0088] S32: Use the preset abnormal damage conditions to judge the abnormal loss data and generate a monitoring result. If the raw material feature similarity in the abnormal loss data is less than the raw material feature similarity threshold in the abnormal damage conditions, and the proportion of the caking area in the abnormal loss data is greater than the caking area proportion threshold in the abnormal damage conditions, the generated monitoring result is to be replenished, and / or if the leakage probability in the abnormal loss data is greater than the leakage probability threshold in the abnormal damage conditions, and the leakage duration in the abnormal loss data is greater than the leakage duration threshold in the abnormal damage conditions, the generated monitoring result is to be replenished; if the raw material feature similarity in the abnormal loss data is greater than or equal to the raw material feature similarity threshold in the abnormal damage conditions, the proportion of the caking area in the abnormal loss data is less than the caking area proportion threshold in the abnormal damage conditions, the leakage probability in the abnormal loss data is less than or equal to the leakage probability threshold in the abnormal damage conditions, and / or the leakage duration in the abnormal loss data is less than the leakage duration threshold in the abnormal damage conditions, execute S33.

[0089] Specifically, the preset abnormal damage conditions include: raw material feature similarity threshold, agglomeration area proportion threshold, leakage probability threshold, and leakage duration threshold.

[0090] The specific values of the raw material feature similarity threshold, agglomeration area proportion threshold, leakage probability threshold, and leakage duration threshold are set according to the actual situation.

[0091] S33: Analyze the predicted inventory data using the dynamically adjusted threshold to generate a monitoring result. If there is a predicted inventory in the predicted inventory data that is less than the dynamically adjusted threshold, the generated monitoring result is to be replenished; if each predicted inventory in the predicted inventory data is greater than or equal to the dynamically adjusted threshold, the generated monitoring result is no replenishment required.

[0092] Furthermore, the expression of the dynamically adjusted threshold is:

[0093]

[0094] where is the nth predicted inventory in the predicted inventory data corresponding dynamically adjusted threshold; T s is the fixed safety threshold; β is the morphological change influence factor; EMA(·) is the exponential smoothing operator; Qwlzs is the physical form conversion coefficient; α sb is the industrial production equipment status coefficient; σ ls is the standard deviation of the raw material consumption in the past f days, where f is a natural number; μ ls is the average raw material consumption in the past f days.

[0095] Specifically, the specific value of T s is set according to the actual situation. The specific value of β is set according to the actual situation. In this application, preferably: β = 0.15. The calculation window of EMA for the dynamically adjusted threshold is set according to the actual situation. In this application, preferably it is 4 hours. The specific value of α sb is set according to the actual situation. In this application, preferably: when the equipment is operating normally, α sb takes 0.8; during equipment maintenance, α sb takes 1.2. The specific value of f is set according to the actual situation. In this application, preferably f = 7. The status of industrial production equipment will affect the raw material consumption data.

[0096] S4: Automatically generate a raw material purchase order and send the raw material purchase order to the enterprise side. After receiving the confirmation information sent by the enterprise side according to the raw material purchase order, broadcast an encryption request to the supplier node to complete the raw material purchase order.

[0097] Further, a raw material procurement order is automatically generated through a blockchain smart contract, and the raw material procurement order is sent to the enterprise side. After receiving the confirmation information sent by the enterprise side according to the raw material procurement order, an encryption request is broadcast to the supplier node to complete the raw material procurement order. However, it is not limited to the blockchain smart contract, and the present application is preferably a blockchain smart contract.

[0098] Further, the blockchain smart contract performs the following operations:

[0099] T1: Verify the legality of the digital certificate of the supplier. If it is legal, execute T2; if it is not legal, end the process and send a risk alert to the enterprise side.

[0100] T2: Compare the raw material specification parameters currently provided by the supplier with the raw material specification parameters of historical purchases through homomorphic encryption technology. If they are the same, execute T3; if they are different, end the process and send a risk alert to the enterprise side.

[0101] Specifically, the raw material specification parameters of historical purchases are obtained through the corrected batch information.

[0102] T3: Automatically release the preset advance payment to the supplier to complete the raw material procurement order.

[0103] Specifically, the specific value of the preset advance payment is set according to the actual situation.

[0104] The beneficial effects achieved by the present application are as follows:

[0105] (1) The artificial intelligence-based monitoring method and system of the present application can improve the accuracy and real-time performance of monitoring results

[0106] (2) The artificial intelligence-based monitoring method and system of the present application can realize real-time monitoring and intelligent replenishment of industrial raw materials through multi-source data fusion, and can solve the problems of monitoring blind spots and insufficient accuracy caused by a single data source.

[0107] (3) The artificial intelligence-based monitoring method and system of the present application solve the adaptability contradiction between the static threshold warning mechanism and the dynamic environmental factors by dynamically adjusting the threshold.

[0108] (4) The artificial intelligence-based monitoring method and system of the present application can avoid the supply chain response bottleneck caused by the lag of manual decision-making and the security risks in cross-system data interaction.

[0109] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the protection scope of the present application is intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application. Obviously, those skilled in the art can make various changes and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the protection of the present application and the scope of equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. An artificial intelligence-based monitoring method, characterized in that, It includes the following steps: S1: Obtain real-time raw material data according to a preset acquisition frequency. The real-time raw material data at least includes: current raw material inventory data, raw material physical form change data, and raw material batch information; S2: Analyze the real-time raw material data to obtain raw material analysis data. The raw material analysis data at least includes: current inventory, predicted inventory data, and abnormal loss data; S3: Analyze the raw material analysis data through preset analysis conditions to generate a monitoring result. The monitoring result is to be replenished or not to be replenished. When the monitoring result is to be replenished, execute S4; when the monitoring result is not to be replenished, execute S1; S4: Automatically generate a raw material purchase order and send the raw material purchase order to the enterprise side. After receiving the confirmation information sent by the enterprise side according to the raw material purchase order, broadcast an encryption request to the supplier node to complete the raw material purchase order.

2. The monitoring method based on artificial intelligence according to claim 1, wherein The sub-steps for analyzing the real-time raw material data to obtain raw material analysis data are as follows: S21: Preprocess the real-time raw material data to obtain preprocessed data. The preprocessed data includes: filtered current inventory, raw material form anomaly data, and corrected batch information; S22: Use the filtered current inventory in the preprocessed data as the current inventory; S23: Construct input variables based on the preprocessed data, temperature parameters, and humidity parameters, and input the input variables into a pre-constructed prediction model. The prediction model outputs the predicted inventory for the next N time instants as the predicted inventory data; S24: Analyze the preprocessed data to obtain abnormal loss data. The abnormal loss data at least includes: raw material feature similarity, caking area ratio, leakage probability, and leakage duration; S25: Use the current inventory, predicted inventory data, and abnormal loss data as the raw material analysis data.

3. The monitoring method based on artificial intelligence according to claim 2, characterized in that, The sub-steps for preprocessing the real-time raw material data to obtain preprocessed data are as follows: S211: Filter the raw material current inventory data in the real-time raw material data through the Kalman filter algorithm to obtain the filtered current inventory; S212: Extract features from the raw material physical form change data in the real-time raw material data through an object detection algorithm to obtain raw material form anomaly data. The raw material form anomaly data at least includes: caking index, leakage probability, looseness, and stacking stability index; S213: Correct the transmission error of the raw material batch information in the real-time raw material data to obtain the corrected batch information; S214: Use the filtered current inventory, raw material form anomaly data, and corrected batch information as the preprocessed data.

4. The monitoring method based on artificial intelligence according to claim 2, characterized in that, The expression of the input variable is: Among them, X tc is the input variable, and t c represents the time point for collecting real-time data of raw materials; Qdqkc is the current inventory after filtering in the preprocessed data; Qwlzs is the physical form conversion coefficient; Pllxh is the theoretical consumption in the next N moments in the production plan scheduling, where N is a natural number; Ewd is the temperature parameter of the raw material storage area; Esd is the humidity parameter of the raw material storage area.

5. The monitoring method based on artificial intelligence according to claim 1, wherein The sub-steps for analyzing the raw material analysis data through preset analysis conditions to generate a monitoring result are as follows: S31: Use a preset inventory safety threshold to judge the current inventory and generate a monitoring result. If the current inventory is less than or equal to the preset inventory safety threshold, the generated monitoring result is to be replenished; if the current inventory is greater than the preset inventory safety threshold, execute S32; S32: Judge the abnormal loss data by using the preset abnormal damage conditions to generate a monitoring result. If the raw material feature similarity in the abnormal loss data is less than the raw material feature similarity threshold in the abnormal damage conditions, and the proportion of the caking area in the abnormal loss data is greater than the caking area proportion threshold in the abnormal damage conditions, the generated monitoring result is to be replenished, and / or if the leakage probability in the abnormal loss data is greater than the leakage probability threshold in the abnormal damage conditions, and the leakage duration in the abnormal loss data is greater than the leakage duration threshold in the abnormal damage conditions, the generated monitoring result is to be replenished; if the raw material feature similarity in the abnormal loss data is greater than or equal to the raw material feature similarity threshold in the abnormal damage conditions, the proportion of the caking area in the abnormal loss data is less than the caking area proportion threshold in the abnormal damage conditions, the leakage probability in the abnormal loss data is less than or equal to the leakage probability threshold in the abnormal damage conditions, and / or the leakage duration in the abnormal loss data is less than the leakage duration threshold in the abnormal damage conditions, then execute S33; S33: Analyze the predicted inventory data by using the dynamically adjusted threshold to generate a monitoring result. If there is a predicted inventory in the predicted inventory data that is less than the dynamically adjusted threshold, the generated monitoring result is to be replenished; if each predicted inventory in the predicted inventory data is greater than or equal to the dynamically adjusted threshold, the generated monitoring result is not to be replenished.

6. The monitoring method based on artificial intelligence according to claim 5, characterized in that, The expression of the dynamically adjusted threshold is: Among them, is the nth predicted inventory in the predicted inventory data corresponding dynamic adjustment threshold; T s is the fixed safety threshold; β is the morphological change influence factor; EMA(·) is the exponential smoothing operator; Qwlzs is the physical form conversion coefficient; α sb is the industrial production equipment status coefficient; σ ls is the standard deviation of raw material consumption in the past f days, and f is a natural number; μ ls is the average raw material consumption in the past f days.

7. The monitoring method based on artificial intelligence according to claim 1, wherein Automatically generate a raw material purchase order through the blockchain smart contract, and send the raw material purchase order to the enterprise side. After receiving the confirmation information sent by the enterprise side according to the raw material purchase order, broadcast an encryption request to the supplier node to complete the raw material purchase order.

8. An artificial intelligence-based monitoring system, characterized in that, Including: A data collection subsystem, an artificial intelligence monitoring center, and an enterprise side; Among them, the data collection subsystem: is used to collect the real-time raw material data according to the preset collection frequency, and send the real-time raw material data to the artificial intelligence monitoring center; The artificial intelligence monitoring center: is used to execute the artificial intelligence-based monitoring method described in any one of claims 1-7; The enterprise side: is used to receive the raw material purchase order and send a confirmation information according to the raw material purchase order.

9. The monitoring system based on artificial intelligence according to claim 8, characterized in that, The data collection subsystem at least includes: a weight sensor, an environmental sensor, an image acquisition device, and an RFID tag reader; Among them, the weight sensor: is used to collect the current inventory data of the raw materials; The environmental sensor: is used to collect the temperature parameter of the raw material storage area and the humidity parameter of the raw material storage area; The image acquisition device: is used to obtain the physical form change data of the raw materials; The RFID tag reader: is used to obtain the raw material batch information.

10. The monitoring system based on artificial intelligence according to claim 9, characterized in that, The image acquisition device is an industrial camera.

Images