Intelligent energy consumption monitoring system and method based on full-level data calibration and dual-dimensional energy consumption benchmarking

Abstract

This invention discloses an intelligent traceability energy consumption monitoring system and method based on full-level data calibration and dual-dimensional energy consumption benchmarking, belonging to the field of energy consumption monitoring technology. It includes a model and authorization layer, a mid-level decision-making layer, a robot edge execution layer, and an industrial equipment layer. The model and authorization layer generates a general model and code verification. The robot edge execution layer uploads a local authorization verification code and verifies it through the model and authorization layer. The robot edge execution layer receives and runs the general model issued by the model and authorization layer and collects the operating data of the industrial equipment layer. The robot edge execution layer analyzes and processes the collected operating data to form analysis results and solutions, which are then uploaded to the mid-level decision-making layer. The mid-level decision-making layer views the solutions and execution results of the robot edge execution layer and adjusts its strategies accordingly. This invention can collect and calibrate the power consumption data of industrial equipment and then provide corresponding energy-saving solutions.

Description

Technical Field

[0001] This invention relates to the field of energy consumption monitoring technology, and in particular to an intelligent traceability energy consumption monitoring system and method based on full-level data calibration and dual-dimensional energy consumption benchmarking. Background Technology

[0002] Industrial equipment is generally old and diverse, consuming a lot of electricity, making unified monitoring and management difficult. Often, equipment operates at high power even when production is low or not in operation, resulting in significant energy waste. Currently, some manufacturers add inspection robots to their factories to periodically monitor equipment energy consumption for better management. However, these existing inspection robots often produce inaccurate data, impacting subsequent decision-making.

[0003] Therefore, to solve the aforementioned technical problems, a new technical solution is needed. Specifically, a smart energy consumption monitoring system and method based on full-level data calibration and dual-dimensional energy consumption benchmarking is required. Summary of the Invention

[0004] To address the technical problem of inaccurate detection data from existing inspection robots, this invention provides an intelligent traceability energy consumption monitoring system and method based on full-level data calibration and dual-dimensional energy consumption benchmarking.

[0005] To achieve the above objectives, the following technical solution is provided: an intelligent traceability energy consumption monitoring system based on full-level data calibration and dual-dimensional energy consumption benchmarking, characterized by: including a model and authorization layer, a mid-level decision-making layer, a robot edge execution layer, and an industrial equipment layer; the model and authorization layer generates a general model and code verification, and pre-issues the authorization verification code and the general lightweight model to the robot edge execution layer for temporary storage to support offline operation; the robot edge execution layer uploads the local authorization verification code and verifies it through the model and authorization layer; the robot edge execution layer receives and runs the general model issued by the model and authorization layer; and collects running data in offline mode; when the model inference error rate of this offline operation is detected to be greater than the pre-defined error rate, the system will detect the error rate of the model inference layer. When a threshold is set, the current model version is uninstalled and automatically rolled back to the previous stable general model. The robot edge execution layer collects the operating data of the industrial equipment layer and performs full-level data calibration on the collected data. The robot edge execution layer analyzes and processes the collected operating data to form analysis results and solutions. The robot edge execution layer generates a visualization report from the operating data, analysis results, and solutions. The robot edge execution layer uploads the operating data, analysis results, and visualization report to the mid-level decision layer. The mid-level decision layer adjusts and executes the parameter configuration of the robot edge execution layer. The mid-level decision layer can query the model and the general model version currently issued by the authorization layer.

[0006] Preferably, the model and authorization layer are cloud devices; the robot edge execution layer includes multiple inspection robot devices; and the mid-level decision-making layer is a computer or tablet.

[0007] Preferably, the system also includes an offline mode: after the authorization verification code of the robot edge execution layer passes, the generated general model and authorization verification code are pre-sent to the robot edge execution layer for temporary storage; the mid-level decision layer sends all the dedicated inspection routes, inspection parameters, inspection tasks, inspection maps, and inspection points to the robot edge execution layer; the robot edge execution layer normally collects the running data and runs the current general model; when the model inference error rate of this offline run is detected to be >5%, the robot edge execution layer uninstalls the current model version, automatically downloads and replaces it with the previous version of the historical stable general model.

[0008] Preferably, the operation steps of the model and the authorization layer include:

[0009] S41. Based on existing equipment electrical characteristics, rated parameters, operating curves, and fault mechanisms, a general model is constructed by combining simulated operating condition data, standard operating condition samples, and industry-standard fault characteristics.

[0010] S42. Establish a cloud-based verification database, which includes robot code, model usage type, and license validity period; establish a cloud-based verification database, which includes robot code, model usage type, and license validity period.

[0011] A cloud-based SN authorization whitelist database is built to record legitimate robot SN codes, hardware fingerprint hash values, authorization validity periods, and model usage permissions; it supports remote permission revocation, and directly blacklists expired, lost, or unsubscribed robots, prohibiting their use.

[0012] S43. Verify that the connected inspection robot equipment code conforms to the general model issued to the inspection robot.

[0013] Preferably, the operation steps of the robot edge execution layer include:

[0014] S51. Encoding Binding and Verification

[0015] The CPU serial number, motherboard MAC address, and inspection module number of the inspection robot in the robot edge execution layer are combined to generate a unique hardware fingerprint and the inspection robot's SN code. The encrypted SN code is then transmitted to the model and authorization layer for verification and encoding.

[0016] S52. System Initialization

[0017] The lightweight model is downloaded sequentially from the model and the authorization layer, and historical versions are stored.

[0018] S53. Multidimensional Data Acquisition and Preprocessing

[0019] The inspection robot collects equipment status data, scene feature data, and cost-related data from the industrial equipment layer.

[0020] The general model preprocesses the collected data, including: 1) integrity verification, triggering re-collection when the data missing rate is >5%; 2) rationality verification, marking as abnormal when the value exceeds the device's rated value by ±30%; 3) time sequence verification, marking and repairing the data by linear interpolation when adjacent data change by >20%.

[0021] S54. Full-level data calibration

[0022] The general model performs instrument status validity verification, multi-source data cross-verification, and historical trend verification on the collected data. Instrument status validity verification includes confirming through image analysis that the instrument is functioning normally, without damage, modification, or abnormal external connections, ensuring the validity of the data foundation. Multi-source data cross-verification involves comparing visual recognition readings, gateway-acquired data, and official platform data; if the error is within 1%–3%, the data is considered accurate. Historical trend verification involves comparing real-time data with historical energy consumption, combined with the reasonable aging range of the equipment, to complete the final verification and output reliable benchmark energy consumption data. Multi-source data cross-verification includes: acquiring instrument image readings through a visual recognition module, acquiring electrical signal data through an industrial gateway, and connecting to official platform data; aligning the three in time and space before comparison; utilizing the non-contact characteristics of visual data to correct the contact measurement error of the gateway data; if the error is within a preset range, the data is considered accurate.

[0023] S55. Automatic Scene Recognition and Dual-Dimensional Benchmarking for High Energy Consumption

[0024] S551. Automatic Scene Recognition

[0025] Based on the completion of data collection and calibration, the general model first automatically identifies the production scenario. The general model has a pre-set basic scenario template. The automatic scenario identification method identifies the scenario from the collected data through rule screening and clustering optimization. After the scenario identification is completed, a unique scenario label is assigned to each time window and the scenario feature vector is associated.

[0026] S552. Dual-Dimensional Intelligent Benchmarking

[0027] When similar equipment exists, its energy consumption is compared with the average energy consumption of equipment in the same workshop, of the same model, in the same scenario, and with the same load rate. If the energy consumption exceeds the preset range, it is judged as high energy consumption. When similar equipment does not exist, its energy consumption is compared with its own historical baseline for the same period, in the same scenario, and under the same working conditions. If the energy consumption exceeds the preset range, it is judged as high energy consumption.

[0028] S56. Root Cause Analysis of Energy Consumption

[0029] The general model first collects equipment status data, scene feature data, cost correlation data, and historical data to form a three-dimensional traceability dataset of "equipment-scene-cost". Then, it establishes eight categories of high energy consumption root cause definitions and judgment rules for scene-cost correlation. Based on the judgment rules, the detected data is automatically classified and traced for root causes.

[0030] S57. Generate an energy-saving solution

[0031] After the general model traces the specific root cause of high energy consumption, it automatically matches the preset "root cause-scenario-cost-measure" association library, and combines real-time data thresholds, scenario constraints, and cost factors to provide a lightweight energy-saving solution that combines emergency handling, optimization and adjustment, and long-term management.

[0032] S58. Energy Saving Quantitative Assessment

[0033] The general model is based on calibrated data and energy-saving solutions. It automatically calculates equipment energy-saving indicators, scenario adaptation indicators, and cost-benefit indicators using preset formulas and generates a visual report.

[0034] S59. Energy Consumption Dual Warning

[0035] When high-energy-consuming equipment is detected, a real-time high-energy-consumption warning is issued. Specifically, an alarm is triggered immediately when the horizontal / vertical benchmark exceeds a predetermined value.

[0036] Preferably, the parameters include voltage, current, power, power factor, load rate, operating time, number of start-stop cycles, temperature, and vibration value.

[0037] Preferably, the production shift, production mode, output data, equipment operating status, ambient temperature and humidity, and air pressure are included.

[0038] Preferably, the cost-related data includes: local preset electricity price, carbon price, and equipment operation and maintenance cost benchmark value.

[0039] This invention also provides an intelligent traceability energy consumption monitoring method based on full-level data calibration and dual-dimensional energy consumption benchmarking, comprising the aforementioned monitoring system. The specific steps are as follows: S91. Generate a general model and an authorized inspection robot device code in the cloud, forming a model and authorization layer; S92. Encode and bind the inspection robot and connect it to the model and authorization layer. After passing the verification by the model and authorization layer, the model and authorization layer activate the general model below the inspection robot to form a robot edge execution layer; S93. The robot edge execution layer performs inspection of the industrial equipment to be inspected according to a predetermined method, including dedicated inspection. The route, inspection parameters, inspection tasks, inspection map, and inspection points can be modified by the mid-level decision-making layer; S94. After the robot edge execution layer inspects the industrial equipment layer, it generates equipment status data. The general model performs data calibration, data analysis, and high-energy-consumption root cause tracing on the equipment status data to generate a solution; The robot edge execution layer generates a visual report from the equipment status data, data analysis results, and solution and uploads it to the mid-level decision-making layer; S95. The mid-level decision-making layer queries the model and the existing general model in the authorization layer, and can adjust and modify the configuration parameters of the robot edge execution layer. Beneficial effects

[0040] 1. Zero modification cost: Reuse the existing data collection and execution modules of the inspection robot, without the need to add additional sensors or modify on-site equipment, significantly reducing the deployment cost and implementation difficulty for enterprises.

[0041] 2. High-precision diagnosis: Through multi-layer data calibration, dual-dimensional intelligent benchmarking and multi-type root cause threshold determination, the root cause of high energy consumption can be accurately located, effectively reducing misjudgment and missed judgment.

[0042] 3. Full-scenario adaptation: Supports offline operation of robots at the edge throughout the entire process, and can adapt to complex working conditions such as poor network or no network in industrial sites; at the same time, it supports unified scheduling and collaborative management of multiple robots to improve operation and maintenance efficiency.

[0043] 4. Forming a commercial closed loop: By adopting dual authorization binding of SN code and hardware fingerprint, software theft is prevented while ensuring normal offline use, and commercial subscription models by device or by year can be supported.

[0044] 5. Energy-saving effects can be quantified and verified: Evaluation indicators such as electricity saving, electricity saving rate, electricity cost savings, and carbon saving are clear and calculable, and energy-saving benefits are quantifiable and traceable, which facilitates enterprise management and decision-making.

[0045] 6. High data security: Core field data is preferentially stored on the robot edge and the manufacturer's central control terminal, and the scope and permissions of data upload are strictly controlled to meet the data security and compliance requirements of industrial scenarios.

[0046] 7. Unified management and strong collaboration: Based on a four-layer architecture, the system enables unified model distribution and centralized policy configuration, ensuring consistent execution standards across multiple robots, thereby improving system versatility and reducing operation and maintenance costs.

[0047] 8. Self-optimization closed-loop capability: Automatic 7-day effect tracking after solution implementation; automatic correction of judgment parameters and scenario coefficients when deviation exceeds threshold, continuously improving diagnostic and energy-saving adaptation accuracy.

[0048] 9. Comprehensive risk warning: It combines real-time high energy consumption alarms with 7-day trend three-level early warning to enable early prediction of energy consumption anomalies and reduce production energy waste and equipment failure risks. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the overall framework of the system of the present invention.

[0050] Figure 2 This is a schematic diagram of the robot edge execution layer initialization process of the system of the present invention.

[0051] Figure 3 This is a schematic diagram of the overall workflow of the system of the present invention.

[0052] Figure 4 This is a schematic diagram of the overall workflow of the system of the present invention when the network is disconnected.

[0053] Figure 5 This is a schematic diagram of the execution steps of the robot edge execution layer in the system of the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are merely illustrative and are not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0055] In the description of this invention, it should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to or indirectly connected to the other element.

[0056] In the description of this invention, it should be noted that the terms "center," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. "Several" means one or more, unless otherwise explicitly specified.

[0057] like Figures 1 to 5As shown, this invention provides an intelligent traceability energy consumption monitoring system based on full-level data calibration and dual-dimensional energy consumption benchmarking. It includes a model and authorization layer, a mid-level decision-making layer, a robot edge execution layer, and an industrial equipment layer. The model and authorization layer generates a general model and code verification. The robot edge execution layer uploads a local authorization verification code and verifies it through the model and authorization layer. The robot edge execution layer receives and runs the general model issued by the model and authorization layer. Before issuance, the model and authorization layer compresses the general model to form a lightweight general model. The robot edge execution layer collects operational data from the industrial equipment layer. The robot edge execution layer analyzes and processes the collected operational data to generate analysis results and solutions. The robot edge execution layer generates a visualization report from the operational data, analysis results, and solutions. The robot edge execution layer uploads the operational data, analysis results, and visualization report to the mid-level decision-making layer. The mid-level decision-making layer adjusts and executes the parameter configuration of the robot edge execution layer. The mid-level decision-making layer can query the current version of the general model issued by the model and authorization layer. The model and authorization layer are cloud devices; the robot edge execution layer includes multiple inspection robot devices; and the mid-level decision-making layer is a computer or tablet.

[0058] The operation steps of the model and the authorization layer include: S41. Based on the existing electrical characteristics, rated parameters, operating curves, and fault mechanisms of the equipment, a general model is constructed by combining simulated operating condition data, standard operating condition samples, and industry-standard fault characteristics.

[0059] The general model is built upon publicly available electrical characteristics, rated parameters, operating curves, and fault mechanisms of equipment such as motors, pumps, fans, and compressors. It combines simulated operating condition data, standard operating condition samples, and industry-standard fault characteristic models, without relying on any manufacturer's actual production data, operating data, or private data. Model components include: full-level data calibration rules, scene recognition algorithms, two-dimensional benchmarking logic, eight types of root cause determination rules, energy-saving solution matching rules, and quantitative evaluation formulas—a black-box model that does not rely on data training.

[0060] S42. Establish a cloud-based verification database, which includes robot codes, model usage types, and license validity periods.

[0061] The specific method is as follows: Whitelist management: Establish a cloud-based SN whitelist library to record the SN code, hardware fingerprint hash value, authorization validity period, and model usage permissions of legitimate robots.

[0062] Authorization verification process: When the robot powers on / loads the model, it initiates an authorization request → The cloud verifies whether the SN code is on the whitelist, whether the authorization has expired, and whether the user has permission to use the model → If the verification passes, a dynamic authorization token is issued.

[0063] Offline authorization fallback: An encrypted offline authorization certificate (including SN, fingerprint hash, and validity period) is issued to the authorized robot. When the network is disconnected, the robot locally verifies the hash consistency. If it passes the verification, the model is temporarily activated.

[0064] Permission revocation: If the robot is lost, the authorization expires, or the subscription is terminated, the SN code will be blacklisted directly in the cloud, prohibiting unauthorized devices from using the model.

[0065] S43. Verify that the connected inspection robot equipment code conforms to the general model issued to the inspection robot.

[0066] Distribution process: Package and encrypt the general model → distribute it to the robot edge via "encrypted transmission + breakpoint resume" → the robot edge receives the data and performs integrity verification → deploy and test the model → if the test fails, automatically roll back to the historical stable version.

[0067] like Figure 4 As shown, this monitoring system is unaffected when the network is disconnected. The following steps are performed in offline mode:

[0068] After the authorization verification code of the robot edge execution layer passes, the generated general model and authorization verification code are pre-sent to the robot edge execution layer for temporary storage.

[0069] The model and authorization layer (offline pre-processing) pre-deploys the general model, version, authorization token, and hardware fingerprint verification rules to the robot's local storage, completes a full authorization verification, and generates a local authorization cache (validity period is configurable, e.g., 7 days). When connected to the network, the robot not only synchronizes data but also automatically reports a statistical report on the inference error rate of this offline run. When offline, the robot's edge execution layer no longer obtains authorization verification from the model and authorization layer but directly verifies through locally stored data.

[0070] The mid-level decision-making layer distributes the dedicated inspection routes, inspection parameters, inspection tasks, inspection maps, and inspection points to the robot's edge execution layer.

[0071] Before going offline, the mid-level decision-making layer sends the execution requirements to the robot's edge execution layer, where the inspection robot completes the task. Data collected offline is not uploaded to the mid-level decision-making layer but is stored locally first. The execution results are uploaded after reconnecting to the network.

[0072] When offline, the robot's edge execution layer is the core layer. This layer normally collects operational data and runs the current general model. When the inference error rate of the model during this offline run exceeds 5%, the edge execution layer uninstalls the current model version, automatically downloads and replaces it with the previous stable general model. Root cause analysis and tracing are performed locally, and the results are temporarily stored. Optimization solutions and evaluation results are generated locally. All data, results, and logs are stored locally and can be viewed on the robot's local screen.

[0073] Existing technologies cannot achieve "dynamic model rollback and authorization verification in the event of network outage", while this solution solves this specific technical problem through "local authorization caching + model version management".

[0074] like Figure 5 As shown, the operation steps of the robot edge execution layer include:

[0075] S51. Encoding Binding and Verification

[0076] The inspection robot's CPU serial number, motherboard MAC address, and inspection module number are combined in the robot edge execution layer to generate a unique hardware fingerprint, which is then encrypted and bound to the inspection robot's serial number (SN). This encrypted SN is then transmitted to the model and authorization layer for verification. The model and authorization layer stores the codes of all legally authorized robots.

[0077] S52. System Initialization

[0078] The lightweight model is downloaded sequentially from the model and the authorization layer, and historical versions are stored.

[0079] like Figure 2 As shown, system initialization includes obtaining a general model version from the cloud (model and authorization layer), and after the model and authorization layer pass verification, the general model is sent to the robot edge execution layer. The robot edge execution layer downloads the latest general model and stores historical versions.

[0080] S53. Multidimensional Data Acquisition and Preprocessing

[0081] The inspection robot collects equipment status data, scene feature data, and cost-related data from the industrial equipment layer.

[0082] Equipment status data includes: voltage, current, power, power factor, load rate, operating time, number of start-stop cycles, temperature, and vibration value.

[0083] Scene feature data includes: production shifts (day shift / night shift), production mode (continuous / intermittent), output data, equipment operating status (no load / half load / full load / maintenance), ambient temperature and humidity, and air pressure.

[0084] Cost-related data includes: local preset electricity price (supports time-of-use pricing), carbon price, and equipment operation and maintenance cost benchmark (single start-up and shutdown cost, aging maintenance cost).

[0085] The general model preprocesses the collected data, including: 1) integrity verification, triggering re-collection when the data missing rate is >5%; 2) rationality verification, marking as abnormal when the value exceeds the device's rated value by ±30%; 3) time sequence verification, marking and repairing the data by linear interpolation when adjacent data change by >20%.

[0086] S54. Full-level data calibration

[0087] The general model performs instrument status validity verification, multi-source data cross-verification, and historical trend verification on the collected data.

[0088] The instrument status validity verification includes confirming through image analysis that the instrument is functioning normally, has not been damaged, modified, or has any abnormal external connections, thus ensuring the validity of the data foundation.

[0089] The multi-source data cross-verification involves comparing visual recognition readings, gateway-collected data, and official platform data. If the error is within 1%–3%, the data is considered accurate.

[0090] The historical trend verification includes comparing real-time data with historical energy consumption, combining it with the reasonable range of equipment aging, completing the final verification, and outputting reliable benchmark energy consumption data.

[0091] Before the general model is enabled, the model and authorization layer will re-verify the general model of the inspection robot by reading the robot's own SN code and verifying whether they match the authorization information (list of allowed SNs, validity period) bound to the model.

[0092] S55. Automatic Scene Recognition and Dual-Dimensional Benchmarking for High Energy Consumption

[0093] S551. Automatic Scene Recognition

[0094] Based on the completion of data collection and calibration, the general model first automatically identifies the production scenario. The general model has a pre-set basic scenario template. The automatic scenario identification method identifies the scenario from the collected data through rule screening and clustering optimization. After the scenario identification is completed, a unique scenario label is assigned to each time window and the scenario feature vector is associated.

[0095] The general model has 5 built-in basic scenario templates: Continuous production scenario: load rate ≥70%, stable output, running time ≥8 hours; Intermittent production scenario: load rate fluctuation >30%, frequent start-stop (≥2 times per hour), output in batches; Night shift low load scenario: load rate <30%, running time between 22:00-6:00, output ≤10% of daily output; Maintenance standby scenario: core equipment shutdown, auxiliary equipment running unloaded, output is 0; Emergency production scenario: load rate >110%, high output priority, running time ≤4 hours.

[0096] The identification method employs a "rule engine + lightweight clustering model," achieving an accuracy of ≥95% through initial rule screening (such as time period + load rate) and cluster optimization.

[0097] S552. Dual-Dimensional Intelligent Benchmarking

[0098] When similar equipment exists, its energy consumption is compared with the average energy consumption of equipment in the same workshop, of the same model, in the same scenario, and with the same load rate. If the energy consumption exceeds the preset range, it is determined to be high energy consumption. When similar equipment does not exist, its energy consumption is compared with its own historical baseline for the same period, in the same scenario, and under the same operating conditions. If the energy consumption exceeds the preset range, it is determined to be high energy consumption. When both dimensions are abnormal or a single dimension exceeds the threshold, the high energy consumption node is identified.

[0099] S56. Root Cause Analysis of Energy Consumption

[0100] The general model first collects equipment status data, scene feature data, cost correlation data, and historical data to form a three-dimensional traceability dataset of "equipment-scene-cost". Then, it establishes eight categories of high-energy-consumption root causes and judgment rules for their correlation with scenes and costs. Based on these judgment rules, the detected data is automatically classified and traced for root causes. The specific classification methods and associated scenes for the eight categories of energy consumption are as follows:

[0101]

[0102] The specific judgment method is as follows: construct a decision tree inference engine based on expert rules, establish a feature factor library containing 8 types of high energy consumption features, scene features, and cost impact, calculate the similarity between real-time data and the feature library through a matching algorithm, and use the decision tree algorithm for weighted scoring. If the total score is ≥80 points, it is determined to be the corresponding root cause.

[0103] S57. Generate an energy-saving solution

[0104] After the general model traces the specific root cause of high energy consumption, it automatically matches the preset "root cause-scenario-cost-measure" association library, and combines real-time data thresholds, scenario constraints, and cost factors to provide a lightweight energy-saving solution that combines emergency handling, optimization and adjustment, and long-term management.

[0105] After the energy consumption data collected by the inspection robot is categorized, the following energy-saving solutions are adopted for the corresponding energy consumption scenarios:

[0106] No-load overtime operation: (1) Emergency period: No-load operation lasts for more than 10 minutes, push down the shutdown reminder (automatic hibernation is triggered in the night shift low load scenario); (2) Optimization plan: Set the no-load overtime threshold according to the scenario (continuous production ≤ 10 minutes, night shift low load ≤ 8 minutes); 3. Post-operation management: Statistical analysis of daily no-load time, triggering team inspection if it exceeds 20%.

[0107] Overload operation: (1) Emergency period: If the energy consumption exceeds the baseline by 15% and it is not an emergency production protection scenario, remind to reduce the load to 90% of the rated load; (2) Optimization plan: Adjust the load threshold according to the aging degree of the equipment, and control the load rate of continuous production scenarios at 80%-95%; (3) Post-production management: Generate an aging trend curve, and replace worn parts before exceeding the baseline by 20%.

[0108] Equipment aging and deterioration: (1) Emergency period: Energy consumption exceeds the baseline by 15%, reminding to shut down for maintenance; (2) Optimization plan: Reduce the load rate of aging equipment (≤80%) and avoid high load operation during peak electricity price periods; (3) Long-term management: Develop a replacement plan according to the aging curve, and prioritize replacement when carbon price is high.

[0109] Unreasonable operating parameters: (1) Emergency period: The parameters deviate from the optimal range of the scenario and are automatically adjusted to the range (e.g., continuous production pressure 0.8-1.0MPa, night shift low load 0.7-0.9MPa); (2) Optimization scheme: The parameters are dynamically adjusted according to the real-time scenario + temperature and humidity; (3) Long-term management: The scenario-parameter template is fixed and automatically called.

[0110] Frequent start-stop operations: (1) Emergency period: ≥3 start-stop operations per hour, lock the equipment, start interval ≥15 minutes; (2) Optimization plan: merge similar operations in intermittent production scenarios and centrally process materials; (3) Long-term management: analyze start-stop patterns, optimize equipment operation sequence, and avoid peak start-stop periods.

[0111] Heat dissipation / insulation failure: (1) Emergency period: The shell temperature exceeds the normal 8℃, reminding you to check heat dissipation / insulation; (2) Optimization plan: In continuous production scenarios, start the backup heat dissipation equipment and wrap the pipes with insulation cotton; (3) Long-term management: Monitor the effect every quarter, and prioritize maintenance when the carbon price is high.

[0112] Interconnection unit coordination imbalance: (1) Emergency period: the main unit is normal and the auxiliary unit is abnormal. Adjust the load of the auxiliary unit to match the main unit (efficiency ≥ 85% in continuous production scenario); (2) Optimization plan: redistribute the unit load according to the scenario. In the night shift low load scenario, shut down the redundant auxiliary unit; (3) Long-term management and control: establish scenario-coordination template and adjust automatically.

[0113] Abnormal leakage / loss: (1) Emergency period: Pressure / flow suddenly drops by more than 10%, reminding to stop the machine for investigation; (2) Optimization plan: Repair the leak point after investigation and replace the seal; (3) Long-term management: Regularly monitor pressure and flow, and give early warning if the pressure and flow drop slowly by more than 5%.

[0114] S58. Energy Saving Quantitative Assessment

[0115] The general model is based on calibrated data and energy-saving solutions. It automatically calculates equipment energy-saving indicators, scenario adaptation indicators, and cost-benefit indicators using preset formulas and generates a visual report.

[0116] S59. Energy Consumption Dual Warning

[0117] When high-energy-consuming equipment is found, a real-time high-energy-consuming warning is issued. Specifically, an alarm is immediately triggered when the horizontal / vertical benchmark exceeds the predetermined value. Dual energy consumption warnings (high-energy-consuming alarm + 7-day trend warning). (1) Real-time high-energy-consuming alarm, Horizontal / vertical benchmarking exceeds ±15% → Immediate alarm. (2) 7-day energy consumption trend early warning, based on historical data of the same period, the same working conditions, the same load, and the same environment, adopts a time series prediction model to predict the energy consumption for the next 7 days on a daily basis. Not limited to 7 days, can be configured by yourself.

[0118] This invention also provides an intelligent energy consumption monitoring method based on full-level data calibration and dual-dimensional energy consumption benchmarking, the specific steps of which are as follows:

[0119] S91. Generate a general model and authorized inspection robot equipment codes in the cloud to form a model and authorization layer.

[0120] S92. Encode and bind the inspection robot and connect it to the model and authorization layer for communication. After the model and authorization layer verify the code, the model and authorization layer enable the general model below the inspection robot to form the robot edge execution layer.

[0121] S93. The robot edge execution layer inspects the industrial equipment to be inspected according to a predetermined method. The predetermined method includes a dedicated inspection route, inspection parameters, inspection tasks, inspection map, and inspection points. The predetermined method can be modified by the mid-level decision-making layer.

[0122] S94. After the robot edge execution layer inspects the industrial equipment layer, it generates equipment status data. The general model performs data calibration, data analysis, and high energy consumption root cause tracing on the equipment status data to generate a solution. The robot edge execution layer generates a visualization report from the equipment status data, data analysis results, and solution and uploads it to the mid-level decision-making layer.

[0123] S95. The intermediate decision layer queries the model and the existing general model of the authorization layer, and can adjust and modify the configuration parameters of the robot edge execution layer.

[0124] This invention provides a monitoring system-level method for collecting, judging, and processing the power consumption of industrial equipment. The system and method can only be used after manual authorization. After collecting and analyzing the data through an inspection robot, a solution is output and displayed on the inspection robot. On-site adjustments are made to achieve energy saving and emission reduction of the equipment.

[0125] The above embodiments are only used to illustrate the present invention and are not intended to limit the present invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all equivalent technical solutions also fall within the scope of the present invention, and the patent protection scope of the present invention should be defined by the claims.

Claims

1. An intelligent energy consumption monitoring system based on full-level data calibration and dual-dimensional energy consumption benchmarking, characterized in that: The system comprises a model and authorization layer, a mid-level decision-making layer, a robot edge execution layer, and an industrial equipment layer. The model and authorization layer generates a general model and code verification. The robot edge execution layer uploads a local authorization verification code and verifies it through the model and authorization layer. The robot edge execution layer receives and runs the general model issued by the model and authorization layer. The robot edge execution layer collects operational data from the industrial equipment layer. The robot edge execution layer analyzes and processes the collected operational data to generate analysis results and solutions. The robot edge execution layer generates a visualization report from the operational data, analysis results, and solutions. The robot edge execution layer uploads the operational data, analysis results, and visualization report to the mid-level decision-making layer. The mid-level decision-making layer adjusts and executes the parameter configuration of the robot edge execution layer. The mid-level decision-making layer can query the currently issued general model version by the model and authorization layer.

2. The energy consumption monitoring system as described in claim 1, characterized in that: The model and authorization layer are cloud devices; the robot edge execution layer includes multiple inspection robot devices; and the mid-level decision-making layer is a computer or tablet.

3. The energy consumption monitoring system as described in claim 1, characterized in that: It also includes an offline mode: after the model and the authorization layer pass the authorization verification code of the robot edge execution layer, the generated general model and authorization verification code are pre-sent to the robot edge execution layer for temporary storage; The mid-level decision-making layer distributes the dedicated inspection route, inspection parameters, inspection tasks, inspection map, and inspection points to the robot's edge execution layer. The robot edge execution layer normally collects the running data and runs the current general model. When it detects that the inference error rate of the model during this offline run is greater than the set value, the robot edge execution layer uninstalls the current model version, automatically downloads and replaces it with the previous version of the historical stable general model.

4. The energy consumption monitoring system as described in claim 1, characterized in that: The operation steps of the model and authorization layer include: S41. Constructing a general model based on existing equipment electrical characteristics, rated parameters, operating curves, and fault mechanisms, combined with simulated operating condition data, standard operating condition samples, and industry-standard fault characteristics; S42. Establishing a cloud-based verification database, which includes robot codes, model usage types, and authorization expiration periods; S43. Verifying that the connected inspection robot equipment code conforms to the specifications before issuing the general model to the inspection robot.

5. The energy consumption monitoring system as described in claim 1, characterized in that: The operation steps of the robot's edge execution layer include: S51. Encoding Binding and Verification The CPU serial number, motherboard MAC address, and inspection module number of the inspection robot in the robot edge execution layer are combined to generate a unique hardware fingerprint and the inspection robot's SN code. The encrypted SN code is then transmitted to the model and authorization layer for verification and encoding. S52. System Initialization Download the lightweight model sequentially from the model and the authorization layer, and store historical versions; S53. Multidimensional Data Acquisition and Preprocessing The inspection robot collects equipment status data, scene feature data, and cost-related data of the industrial equipment layer. The general model preprocesses the collected data, including: 1) integrity verification, triggering re-collection when the data missing rate is greater than a preset value; 2) rationality verification, marking as abnormal when the value exceeds the preset range of the equipment's rated value; 3) time sequence verification, marking and repairing adjacent data changes that are greater than a preset value through linear interpolation. S54. Full-level data calibration The general model performs instrument status validity verification, multi-source data cross-verification, and historical trend verification on the collected data. The instrument status validity verification includes confirming through image analysis that the instrument is functioning normally, has not been damaged, modified, or has any abnormal external connections, ensuring the validity of the data foundation; The multi-source data cross-verification includes comparing visual recognition readings, gateway-collected data, and official platform data. If the error is within a preset range, the data is determined to be accurate. The historical trend verification includes comparing real-time data with historical energy consumption, combining it with the reasonable range of equipment aging, completing the final verification, and outputting reliable benchmark energy consumption data. S55. Automatic Scene Recognition and Dual-Dimensional Benchmarking for High Energy Consumption S551. Automatic Scene Recognition Based on data collection and calibration, the general model first automatically identifies production scenarios. The general model has a pre-set basic scenario template. The automatic scenario identification method uses rule screening and clustering optimization to identify scenarios from the collected data. After the scenario identification is completed, a unique scenario label is assigned to each time window and associated with the scenario feature vector. S552. Dual-Dimensional Intelligent Benchmarking When there are similar devices, their average energy consumption is compared with that of devices in the same workshop, of the same model, in the same scenario, and with the same load rate. If the energy consumption exceeds the preset range, it is judged as high energy consumption. When no similar equipment is available, the system compares the energy consumption with its own historical baseline for the same period, scenario, and working conditions. If the energy consumption exceeds the preset range, it is judged as high energy consumption. S56. Root Cause Analysis of Energy Consumption The general model first collects equipment status data, scene feature data, cost correlation data, and historical data to form a three-dimensional traceability dataset of "equipment-scene-cost"; then it establishes eight categories of high energy consumption root cause definitions and judgment rules for scene-cost correlation; and automatically classifies and traces the root causes of the detected data according to the judgment rules. S57. Generate an energy-saving solution After the general model traces the specific root cause of high energy consumption, it automatically matches the preset "root cause-scenario-cost-measure" association library, and combines real-time data thresholds, scenario constraints, and cost factors to provide a lightweight energy-saving solution that combines emergency handling, optimization and adjustment, and long-term management. S58. Energy Saving Quantitative Assessment The general model is based on calibrated data and energy-saving solutions. It automatically calculates equipment energy-saving indicators, scenario adaptation indicators, and cost-benefit indicators using preset formulas and generates a visual report. S59. Energy Consumption Dual Warning When high-energy-consuming equipment is detected, a real-time high-energy-consumption warning is issued. Specifically, an alarm is triggered immediately when the horizontal / vertical benchmark exceeds a predetermined value.

6. The energy consumption monitoring system as described in claim 5, characterized in that: The equipment status data includes: voltage, current, power, power factor, load rate, operating time, number of start-stop cycles, temperature, and vibration value.

7. The energy consumption monitoring system as described in claim 5, characterized in that: The scene feature data includes: production shifts, production modes, output data, equipment operating status, ambient temperature and humidity, and air pressure.

8. The energy consumption monitoring system as described in claim 5, characterized in that: The cost-related data includes: local preset electricity price, carbon price, and equipment operation and maintenance cost benchmark value.

9. An intelligent energy consumption monitoring method based on full-level data calibration and dual-dimensional energy consumption benchmarking, characterized in that: The system comprising any one of claims 1 to 8, wherein the specific steps are as follows: S91. Generate a general model and authorized inspection robot equipment codes in the cloud to form a model and authorization layer; S92. Encode and bind the inspection robot and connect it to the model and authorization layer for communication. After the model and authorization layer passes the verification, the model and authorization layer enables the general model below the inspection robot to form the robot edge execution layer. S93. The robot edge execution layer inspects the industrial equipment to be inspected according to a predetermined method. The predetermined method includes a dedicated inspection route, inspection parameters, inspection tasks, inspection map, and inspection points. The predetermined method can be modified by the mid-level decision layer. S94. After the robot edge execution layer inspects the industrial equipment layer, it generates equipment status data. The general model performs data calibration, data analysis, and high energy consumption root cause tracing on the equipment status data to generate a solution. The robot edge execution layer generates a visual report from the equipment status data, data analysis results, and solution and uploads it to the mid-level decision-making layer. S95. The intermediate decision layer queries the model and the existing general model of the authorization layer, and can adjust and modify the configuration parameters of the robot edge execution layer.

Images