Tobacco warehouse motor distributed monitoring control method based on loRa transparent polling

Abstract

The application relates to the technical field of tobacco storage environment monitoring and control, in particular to a tobacco warehouse motor distributed monitoring and control method based on LoRa transparent transmission polling. The method comprises the following steps: acquiring a tobacco warehouse multi-parameter information set; performing multi-type fine classification on abnormal behaviors based on the tobacco warehouse multi-parameter information set, to generate an abnormal classification information set; performing dynamic evaluation on a distributed influence domain in combination with warehouse plan and point data correlation, to generate a distributed influence domain evaluation information set; performing regional adaptive linkage control for different risk levels based on the evaluation information set, to generate a motor monitoring and control response information set. The application can effectively solve the problems of high false alarm rate, invisible risk space propagation and control response lag in the prior art, and realizes accurate monitoring and safe closed-loop management and control of the tobacco warehouse motor.

Description

Technical Field

[0001] This application relates to the field of tobacco storage environment monitoring and control technology, and in particular to a distributed monitoring and control method for tobacco warehouse motors based on LoRa transparent polling. Background Technology

[0002] Tobacco warehouses, as crucial sites for tobacco leaf aging and mold prevention, have extremely stringent requirements for the stability of their internal temperature and humidity environments. They typically require a large number of distributed air conditioning units to maintain constant storage conditions. Current technical solutions for monitoring and controlling these air conditioning motors mostly employ wired bus connections or single threshold alarm methods. Wired solutions connect each motor controller to a central monitoring system via cables, collecting basic parameters such as voltage, current, and temperature in real time. Wireless solutions utilize short-range communication modules to upload data to a gateway. When a monitored parameter exceeds a preset fixed threshold, the system triggers an alarm signal, prompting operators to inspect or intervene manually. Some systems can also remotely start and stop individual devices based on commands.

[0003] However, existing technologies lack in-depth analysis of multi-parameter correlation trends and integration with the physical characteristics of warehouse space, making it difficult to effectively distinguish between normal operating condition fluctuations and real safety hazards. This results in a high false alarm rate and an inability to identify the spatial propagation trend of anomalies. Furthermore, existing technologies often rely on manual intervention after detecting single-point anomalies, lacking a risk-level-based regional adaptive linkage mechanism. This makes it difficult to automatically execute targeted, tiered control strategies in the early stages of risk spread, leading to insufficient ability to contain cascading risks. Summary of the Invention

[0004] This application provides a distributed monitoring and control method for motors in tobacco warehouses based on LoRa transparent polling to solve the above problems. The method includes:

[0005] Obtain a multi-parameter information set for the tobacco warehouse, which includes motor active power, motor surface temperature, air outlet temperature, and air outlet humidity.

[0006] Based on the multi-parameter information set of the tobacco warehouse, a refined classification of abnormal behaviors matching the tobacco storage environment is performed, generating an abnormal classification information set including abnormal tags such as heat dissipation obstruction type, condensation and water separation type, and system coupling failure type.

[0007] Based on the aforementioned anomaly classification information set, combined with the tobacco warehouse floor plan and location data correlation, a distributed impact domain dynamic assessment is conducted to generate a distributed impact domain assessment information set containing a comprehensive risk index and impact domain boundary information.

[0008] Based on the distributed influence domain assessment information set, regional adaptive linkage control is executed for different risk levels, generating a motor monitoring and control response information set containing control command sequences and execution status.

[0009] In one optional embodiment, the process of acquiring the multi-parameter information set of the tobacco warehouse includes:

[0010] The SPI interface built into the MCU core control module of each terminal device is used to communicate with the BL0942 power metering chip to read the active power of the motor in real time.

[0011] Simultaneously, through a universal input / output interface, data is read from the motor surface temperature and humidity sensor, the air outlet temperature and humidity sensor, and the smoke sensor, respectively, to collect the motor surface temperature, the air outlet temperature, and the air outlet humidity;

[0012] The MCU core control module assembles the raw data into a response message according to the MODBUS protocol format, transmits it to the LoRa wireless transparent transmission module through the optically isolated serial port unit, and then transmits it back to the host computer via the LoRa gateway and Ethernet, thus summarizing and forming a multi-parameter information set covering the entire tobacco warehouse.

[0013] In one optional embodiment, the LoRa wireless pass-through module employs a polling mechanism based on time slot allocation when performing data backhaul:

[0014] The host computer sends query commands to each terminal device in sequence at a preset polling cycle. After receiving the command corresponding to its own address, each terminal device only starts the LoRa module to respond and upload within the allocated time slot. All terminal devices that do not receive query commands remain in a low-power sleep state.

[0015] In one optional embodiment, the process of generating the anomaly classification information set includes:

[0016] Establish sliding time series with a length of several polling periods for the active power of the motor, the surface temperature of the motor, the air outlet temperature, and the air outlet humidity.

[0017] For each sliding time series, the fluctuation direction and fluctuation amplitude of adjacent periods are calculated to obtain the fluctuation trend vector of each parameter. The four fluctuation trend vectors are cross-compared pairwise to identify synchronous fluctuation patterns and asynchronous fluctuation patterns.

[0018] Based on the synchronous fluctuation mode and the asynchronous fluctuation mode, the abnormal state is classified into the heat dissipation obstruction type, the condensation and water separation type and the system coupling failure type, and an abnormal classification information set is generated for each abnormal point.

[0019] In an optional embodiment, the process for determining the heat dissipation obstruction type includes:

[0020] Extract the fluctuation trend vector of the active power of the motor and the fluctuation trend vector of the surface temperature of the motor, and calculate the trend consistency rate of the two. The trend consistency rate is defined as the proportion of the number of cycles in which the two fluctuate in the same direction to the total number of cycles.

[0021] When the trend consistency rate exceeds the preset trend threshold, and at the same time the fluctuation trend vector of the air outlet temperature shows a continuous decrease, while the fluctuation amplitude of the motor surface temperature shows an accelerating increase, the current anomaly is classified as the heat dissipation obstruction type.

[0022] In an optional embodiment, the process for determining the condensate precipitation type includes:

[0023] Extract the fluctuation trend vector of the air outlet temperature and the fluctuation trend vector of the air outlet humidity, and calculate the number of times the fluctuation direction reverses. The number of fluctuation direction reversals is defined as the number of times the temperature drop segment and the humidity rise segment alternate within a time window.

[0024] When the number of reversals in the direction of fluctuation exceeds a preset reversal threshold, and the time difference between the start time of the temperature drop and the start time of the humidity rise in the reversal event is less than a preset time difference threshold, and the fluctuation trend vectors of the motor's active power and the motor's surface temperature both show stable or random fluctuations, then the current anomaly is classified as the condensation and water separation type.

[0025] In an optional embodiment, the system coupling fault determination process includes:

[0026] Calculate the first correlation between the active power of the motor and the surface temperature of the motor, the second correlation between the outlet temperature and the outlet humidity, and the third correlation between the surface temperature of the motor and the outlet temperature;

[0027] The correlation is defined as the weighted sum of the square of the rate of consistency of the fluctuation direction of the two parameters within the sliding time window and the reciprocal of the difference in fluctuation amplitude.

[0028] When any two of the first correlation degree, the second correlation degree, and the third correlation degree simultaneously deviate from their respective historical normal ranges by more than a preset correlation degree threshold, and the deviation state is maintained continuously for several polling cycles, the current anomaly is classified as the system coupling failure type.

[0029] In one optional embodiment, the process of generating the distributed influence domain assessment information set includes:

[0030] Based on the coordinates of each abnormal point in the abnormal classification information set, retrieve the tobacco warehouse floor plan and identify other terminals in the vicinity of this point.

[0031] Extract the changes in air outlet humidity and motor surface temperature of each terminal in the neighboring area within the most recent polling cycles, and count the number of terminals that show significant changes in the same direction.

[0032] Based on the number of terminals, and combined with the airflow arrow directions marked on the tobacco warehouse floor plan and the warehouse partition locations, the abnormal diffusion direction and the boundary of the affected area are determined.

[0033] Calculate the comprehensive risk index of all terminals within the boundary of the influence domain, and summarize the anomaly classification information and influence domain boundary information to generate the distributed influence domain assessment information set.

[0034] In one optional embodiment, the calculation process of the comprehensive risk index includes:

[0035] Based on the anomaly type of each anomaly point in the anomaly classification information set, a preset anomaly type weight table is retrieved, and a first weight value is assigned to this anomaly point.

[0036] Based on the boundary of the influence domain to which this point belongs, retrieve the preset regional weight table and assign a second weight value to this region;

[0037] Obtain the stock value coefficient of tobacco storage area within the boundary of the influence domain, normalized to the interval [0,1].

[0038] The comprehensive risk index is obtained by multiplying the first weight value, the second weight value, and the stock value coefficient.

[0039] In one optional embodiment, the process of generating the motor monitoring and control response information set includes:

[0040] Based on the comprehensive risk index in the distributed impact domain assessment information set, priority load reduction or shutdown control commands are executed for the areas where abnormal points at the emergency response level are located.

[0041] For areas not at the emergency response level, implement power-limited operation or ventilation adjustment instructions;

[0042] The sending time, target terminal address, and execution result status of each control command are recorded and summarized to form the motor monitoring and control response information set.

[0043] This application provides a distributed monitoring and control method for motors in tobacco warehouses based on LoRa transparent polling. This method acquires a multi-parameter information set of the tobacco warehouse, including motor active power, motor surface temperature, outlet temperature, and outlet humidity. Based on this, it performs refined classification of abnormal behaviors that match the tobacco storage environment, generating an anomaly classification information set including anomaly tags such as heat dissipation obstruction, condensation, and system coupling failure. Then, it combines the tobacco warehouse floor plan and the correlation of point data to perform a distributed impact domain dynamic assessment, generating a distributed impact domain assessment information set including a comprehensive risk index and impact domain boundary information. Finally, based on this assessment information set, it executes regional adaptive linkage control for different risk levels, generating a motor monitoring and control response information set including control command sequences and execution status. By constructing a multi-parameter sliding time series and calculating the fluctuation trend vector, and utilizing a synchronous and asynchronous fluctuation pattern recognition mechanism, the system can accurately distinguish abnormal states caused by different factors, such as heat dissipation obstruction, condensation and water precipitation, and system coupling failures. This effectively solves the problems of high false alarm rates and inability to identify abnormal types caused by traditional single threshold judgment. On this basis, by using the fusion analysis of the warehouse planar topology and the correlation of data from adjacent terminals, the influence domain boundary of abnormal diffusion is dynamically delineated and the comprehensive risk index is quantified. This avoids the limitations of invisible risk spatial propagation paths and reliance on manual experience to judge the scope of influence. Furthermore, based on the quantified risk level, a hierarchical linkage control strategy is automatically triggered, ranging from routine adjustments to zonal isolation and even full warehouse alert. This achieves automated and regional closed-loop handling of potential risks, improving the accuracy of tobacco warehouse motor operation monitoring, the foresight of risk warnings, and the timeliness and reliability of emergency response.

[0044] In summary, this application, through a technical closed loop of multi-dimensional perception, intelligent classification, spatial assessment, and adaptive control, not only achieves refined management and control of the operating status of motors in tobacco warehouses, but also constructs a risk defense system based on a combination of data-driven and physical models, ensuring the safety and stability of the storage environment and the quality of tobacco leaf storage. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic diagram illustrating an application scenario provided in one embodiment of this application;

[0047] Figure 2A flowchart of a distributed monitoring and control method for tobacco warehouse motors based on LoRa transparent polling, provided as an embodiment of this application. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0049] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0050] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0051] Figure 1 This application provides an schematic diagram of an application scenario. In the process of monitoring and controlling motors in tobacco warehouses, the method provided in this application can effectively solve the problems of high false alarm rate, invisible risk space propagation, and delayed control response in the prior art, and realize accurate monitoring and safe closed-loop management of motors in tobacco warehouses.

[0052] Specifically, the method provided in this application can be applied to any server, where the server interacts with the tobacco warehouse online terminal device to obtain the multi-parameter information set of the tobacco warehouse provided by the tobacco warehouse online terminal device, and finally outputs the motor monitoring and control response information set to relevant personnel.

[0053] The specific implementation method can be referred to in the following embodiments, wherein the data mentioned in the embodiments are only for reference and examples, so that relevant personnel can better understand them.

[0054] Figure 2 This is a flowchart illustrating a distributed monitoring and control method for tobacco warehouse motors based on LoRa transparent polling, provided in one embodiment of this application. The method of this embodiment can be applied to the server in the above scenario. (See also...) Figure 2 The process, specifically the implementation steps, are as follows:

[0055] Example 1:

[0056] Tobacco warehouses have extremely stringent requirements for temperature and humidity conditions, and are typically equipped with a large number of air conditioning units to maintain the stable conditions needed for tobacco aging and mold prevention. However, existing monitoring methods mostly use wired buses or single threshold alarms, which have problems such as data silos making it impossible to identify the spatial propagation trend of anomalies, simple threshold judgments confusing normal operating fluctuations with real hidden dangers leading to a high false alarm rate, and reliance on manual handling after the discovery of a single point of anomaly, resulting in delayed control response and a lack of regional linkage. These issues make it difficult to effectively curb the chain risks such as the spread of local heat sources or humidity diffusion.

[0057] To address the aforementioned issues, this application provides a distributed monitoring and control method for tobacco warehouse motors based on LoRa transparent polling. The method includes:

[0058] Step 1: Obtain the multi-parameter information set of the tobacco warehouse, which includes the active power of the motor, the surface temperature of the motor, the temperature of the air outlet, and the humidity of the air outlet.

[0059] The multi-parameter information set for the tobacco warehouse refers to the collection of real-time operational data collected from all online terminal devices covering the entire warehouse. This data is obtained synchronously from terminal devices deployed next to each air conditioner motor, utilizing built-in energy metering chips and multi-sensor modules. Specifically, the motor's active power reflects its load status and energy consumption level; the motor's surface temperature directly characterizes the heat accumulation within the motor; and the outlet temperature and humidity jointly reflect the cooling and dehumidification efficiency of the air conditioning system and the heat and humidity exchange state of the surrounding air. These four parameters are not isolated but are linked through time series to form a multi-dimensional feature vector for subsequent in-depth analysis. For example, when the motor is operating under high load, an increase in active power is usually accompanied by an increase in surface temperature, but if the outlet temperature does not change accordingly, it suggests a possible blockage in the heat dissipation path. By simultaneously acquiring these four types of parameters, a comprehensive perception of the motor's operating status can be built, providing sufficient data support for distinguishing between normal fluctuations and actual faults. This step aims to break down the data silos of traditional single-point monitoring, aggregating scattered physical quantities into a unified digital foundation, thereby providing complete input conditions for subsequent refined classification.

[0060] Step 2: Based on the multi-parameter information set of the tobacco warehouse, perform multi-type refined classification of abnormal behaviors that match the tobacco storage environment, and generate an abnormal classification information set that includes abnormal tags such as heat dissipation obstruction type, condensation and water separation type, and system coupling failure type.

[0061] The anomaly classification information set is a structured data record generated by combining and analyzing the changing trends, fluctuation amplitudes, and interrelationships of various parameters in the multi-parameter information set. Its generation process is customized based on the unique physical environmental characteristics of tobacco storage (such as flammable dust accumulation and the risk of mold growth in high humidity), rather than a simple judgment of numerical limits exceeding limits. The heat dissipation obstruction anomaly tag indicates a state where heat cannot be dissipated in a timely manner due to filter blockage or fan failure. Its characteristic is that the motor's active power and surface temperature rise synchronously, while the outlet temperature remains relatively stable or decreases. The condensation and water precipitation anomaly tag indicates a state where humidity near the outlet rises sharply due to excessive dehumidification of the refrigeration system or refrigerant leakage. Its characteristic is a sudden drop in outlet temperature accompanied by a rapid increase in humidity, while the motor's parameters show no significant change. The system coupling failure anomaly tag indicates a state where comprehensive problems occur within the equipment. Its characteristic is that three or more core data points simultaneously exhibit abnormal fluctuations exceeding historical statistical baselines and persist for multiple cycles. For example, if the system detects that the active power at a certain point increases from 2.5kW to 3.2kW, the surface temperature rises from 36℃ to 48℃, and the outlet temperature drops from 16℃ to 14℃, the system will identify the positive correlation between power and temperature and the inverse correlation between the outlet temperature through trend vector comparison, thus determining it as a heat dissipation obstruction type. This multi-type refined classification effectively eliminates interference from normal operating condition fluctuations caused by load switching, significantly reducing the false alarm rate and enabling operators to accurately locate the type of potential hazard.

[0062] Step 3: Based on the anomaly classification information set, combined with the tobacco warehouse floor plan and the correlation of location data, conduct a distributed impact domain dynamic assessment to generate a distributed impact domain assessment information set containing a comprehensive risk index and impact domain boundary information;

[0063] The distributed impact domain assessment information set is the result of mapping abstract anomaly data to a specific physical space for risk quantification. Its generation relies on a pre-constructed tobacco warehouse floor plan (including equipment coordinates, shelf distribution, and airflow direction) and data correlation analysis of adjacent terminals. The comprehensive risk index is the core indicator for quantifying the hazard level of anomaly locations. It is calculated by weighting the inherent hazard weight of the anomaly type, the spatial diffusion range level of the anomaly, and the value coefficient of tobacco stock within the impact domain. The impact domain boundary information defines the physical range that the anomaly may affect. It is determined by extracting parameter changes from other terminals in the vicinity of the anomaly location, counting the number of terminals showing significant changes in the same direction, and dynamically correcting this by combining the airflow arrow direction marked on the floor plan with the warehouse partition location. For example, if a location is identified as a condensation anomaly, and the humidity at the air outlets of three adjacent locations within a 12-meter downwind radius simultaneously increases by more than 5%, the system will designate this area as a high-risk impact domain and assign it a higher comprehensive risk index. This step combines data correlation analysis with spatial topology to upgrade risk warning from point-based to spatialized, enabling the system to predict the spread path of anomalies and the potential range of hazards.

[0064] Step 4: Based on the distributed impact domain assessment information set, execute regional adaptive linkage control for different risk levels to generate a motor monitoring and control response information set containing control command sequences and execution status.

[0065] The motor monitoring and control response information set is a closed-loop feedback record after the system executes automated decisions, containing the specific control command sequence and the execution result status of each terminal. Regional adaptive linkage control refers to automatically matching different levels of handling strategies based on the magnitude of the comprehensive risk index and the type of anomaly, executing graded actions from enhanced monitoring to emergency shutdown without manual intervention. Specifically, for low-risk levels, the control strategy may simply be to shorten the polling interval to increase monitoring density; for medium-risk levels, load reduction or intermittent operation commands are sent to the anomaly location and terminals within its affected area; for high-risk levels, a zoned emergency shutdown protocol is triggered, forcibly cutting off power and locking the remote restart function. For example, when the comprehensive risk index exceeds 0.6 and humidity diffusion characteristics are identified, the system will automatically generate and send an emergency shutdown command to all terminals within the affected area, recording the command sending time, target address, and confirmation receipt. Through this graded linkage mechanism, the system can block the spread of risk at the first moment of a fault, transforming traditional passive manual handling into proactive automated closed-loop control, significantly improving the security capabilities of tobacco warehouses.

[0066] This application constructs a complete closed loop of perception, cognition, assessment, and decision-making through the synergistic effect of the aforementioned technical features. First, the acquisition of a multi-parameter information set provides the system with comprehensive basic perception capabilities, solving the problem of biased judgment caused by data gaps. Based on this, the refined classification of multiple types of abnormal behaviors matched to the tobacco storage environment utilizes the logical combination of multi-dimensional parameters to achieve accurate recognition of real potential hazards and effectively filter out invalid alarms. Furthermore, the distributed impact domain dynamic assessment introduces spatial topology and data correlation, expanding isolated anomalies into visualized risk areas and quantifying the risk propagation trend. Finally, regional adaptive linkage control based on risk assessment results automatically executes differentiated strategies according to risk levels, achieving a rapid response from anomaly detection to risk prevention. This progressively layered technical solution not only reduces the false alarm rate but also solves the problems of invisible risk spatial propagation and delayed control response in existing technologies through spatial early warning and automated linkage, ensuring the safety and stability of the tobacco warehouse storage environment.

[0067] Example 2:

[0068] In one optional embodiment, this application provides a specific implementation method for obtaining a multi-parameter information set of a tobacco warehouse.

[0069] Step 1: Communicate with the BL0942 energy metering chip through the SPI interface built into the MCU core control module of each terminal device to read the active power of the motor in real time;

[0070] The MCU core control module refers to the microprocessor unit deployed in the terminal device next to each air conditioner motor. Specifically, it can use an industrial-grade 32-bit chip such as the STM32F103C8T6, which integrates a Serial Peripheral Interface (SPI) for high-speed synchronous data communication. The BL0942 energy metering chip is a high-precision dedicated integrated circuit for energy metering. It collects the voltage signal of the motor circuit through a precision voltage divider resistor network and collects the current signal through a manganese-copper shunt resistor sampling circuit. The MCU core control module sends a read command to the BL0942 via the SPI interface. The BL0942 returns the internally calculated register data such as voltage, apparent power, active power, and power factor to the MCU serially. This step directly extracts the core electrical parameters reflecting the motor load status from the high-voltage circuit, ensuring the real-time performance and accuracy of the data. For example, when the motor is running under rated load, the BL0942 chip performs high-speed sampling and DSP calculations on the voltage and current waveforms through its internal ADC, calculates the real-time active power as 3.2kW, and transmits this value in 32-bit floating-point format to the MCU's receive buffer via the SPI bus. This layout utilizes the central region as an electromagnetic buffer zone, effectively blocking interference from the high-voltage area at the top to the digital signals in the low-voltage area at the bottom, thus ensuring the stability of the SPI communication link. Direct communication between the MCU and the metering chip eliminates errors introduced by intermediate conversion stages, providing highly reliable power reference data for subsequent abnormal behavior classification.

[0071] Step 2: Simultaneously, through the general input / output interface, data is read from the motor surface temperature and humidity sensor, the air outlet temperature and humidity sensor, and the smoke sensor to collect the motor surface temperature, air outlet temperature, and air outlet humidity.

[0072] The General Purpose Input / Output Interface (GPIO) is the physical port used by the MCU core control module to communicate with external low-speed sensing devices. The motor surface temperature and humidity sensor and the air outlet temperature and humidity sensor are typically digital sensing elements. Their data communication terminals are connected to the MCU's GPIO port, reading temperature and humidity values ​​via a single bus or I2C protocol. The smoke sensor is typically an analog output or switch output device. Its analog output terminal is connected to the MCU's analog-to-digital converter (ADC) channel via an RC filter circuit, or the alarm status signal is connected to the GPIO interrupt pin. This step aims to simultaneously acquire multi-dimensional physical quantities reflecting the thermal and humidity conditions of the tobacco storage environment and the safety status of the equipment. For example, the MCU polls the GPIO status at preset intervals, reading that the motor surface temperature is 48°C, the air outlet temperature is 14°C, and the air outlet humidity is 72%RH, while the smoke sensor output indicates a smoke-free state. These data are strictly aligned with the acquired active power data in terms of timestamps, forming a complete single-point multi-dimensional sensing data packet. By acquiring data from multiple sensors in parallel, the spatiotemporal correlation between the motor's heating and changes in the exhaust environment can be captured, providing the necessary input conditions for identifying specific abnormal patterns such as obstructed heat dissipation or condensation.

[0073] Step 3: The MCU core control module assembles the raw data into a response message according to the MODBUS protocol format, transmits it to the LoRa wireless transparent transmission module through the optocoupler isolated serial port unit, and then transmits it back to the host computer via the LoRa gateway and Ethernet, thus summarizing and forming a multi-parameter information set covering the entire tobacco warehouse.

[0074] The MODBUS protocol is a widely used master-slave communication protocol in industrial settings. The MCU core control module encapsulates the raw data collected, such as active power, temperature, humidity, and smoke conditions, into a standard MODBUS-RTU frame format according to a predefined register address mapping table. This frame includes an address code, function code, data area, and cyclic redundancy check (CRC) code. An optocoupler-isolated serial port unit is located between the MCU and the LoRa wireless transparent transmission module. It utilizes an optocoupler to achieve contactless transmission of electrical signals, eliminating ground loop interference. The LoRa wireless transparent transmission module operates in transparent transmission mode, modulating the received serial port data into an RF signal and sending it to the LoRa gateway. The LoRa gateway then forwards the data to the host computer system within the local area network via an Ethernet interface. This process achieves end-to-end transmission from local data acquisition to central aggregation, ensuring data integrity and reliability in complex electromagnetic environments. For example, the MCU assembles the collected data into a response message containing address 0x12, function code 0x03, and the corresponding data length. After optocoupler isolation, this message is sent to the LoRa module and wirelessly transmitted to the host computer. By combining optical isolation with LoRa wireless transmission, not only are the high costs and difficult construction of traditional wired cabling solved, but communication packet loss caused by electromagnetic interference from high-voltage circuits is also completely avoided. This enables the host computer to stably acquire a standardized multi-parameter information set covering the entire warehouse, laying a solid data foundation for subsequent refined classification and linkage control.

[0075] This application constructs a layered, decoupled, and highly reliable data acquisition pipeline through the synergistic effect of the aforementioned technical features. The MCU core control module acts as the central hub, ensuring the accuracy of active power metering through a direct SPI interface connection to the BL0942 chip, and ensuring the synchronization of sensing by parallel acquisition of data from multiple environmental sensors via GPIO. Furthermore, by standardizing data encapsulation using the MODBUS protocol, seamless compatibility with existing industrial management platforms is achieved. Moreover, a dual-level isolated wireless link constructed using an optocoupler-isolated serial port unit and a LoRa wireless transparent transmission module effectively blocks strong electrical interference and overcomes spatial wiring limitations. This organic combination of hardware architecture and communication protocol enables dozens or even hundreds of motor terminals distributed throughout the tobacco warehouse to aggregate high-quality electrical and environmental parameters to the host computer with low latency and high consistency, forming a multi-parameter information set covering the entire warehouse. This improves the anti-interference capability, deployment flexibility, and data integrity of the distributed monitoring system, solving the engineering problems of signal susceptibility to interference, protocol incompatibility, and unreliable links in traditional solutions.

[0076] Example 3:

[0077] In one alternative implementation, the method includes a time-slot-based polling mechanism to optimize the data backhaul process.

[0078] Step 1: The host computer sends query commands to each terminal device sequentially at a preset polling cycle;

[0079] The preset polling period refers to the time window required for the host computer to complete a full data acquisition from all online terminal devices. Its value can be dynamically set based on the total number of terminal nodes in the tobacco warehouse and the duration of a single communication. For example, in a scenario supporting a network of 256 terminals, if the average communication time per terminal is 50 milliseconds, the preset polling period can be set to 15 to 20 seconds to balance data real-time performance and channel load. The query command is a data packet containing the unique address identifier of the target terminal. The host computer, as the master station, generates and issues this command sequentially according to the address index order or a preset priority list. This orderly command issuance mechanism ensures that only one valid downlink control signal exists in the network at any given time, avoiding command conflicts caused by multiple master station contention and establishing a time reference for the orderly response of subsequent terminals.

[0080] Step 2: After receiving the instruction corresponding to its own address, each terminal device will only activate the LoRa module to upload the response within the allocated time slot;

[0081] The allocated time slots can be pre-planned or dynamically negotiated time segments dedicated to uplink data transmission for a specific terminal device. When the MCU core control module of the terminal device resolves that the address in the query command matches the address configured by the local device via a DIP switch, it immediately wakes up from low-power sleep mode and activates the LoRa wireless transparent transmission module to enter transmit mode. The length of this time slot is determined based on the size of the response message and the LoRa spreading factor, for example, set to 100 milliseconds. The terminal strictly sends out the assembled MODBUS response message within this time window. The LoRa module works in conjunction with the polling logic of the host computer to completely isolate the uplink signals of different terminals in the time domain by dividing the continuous time axis into discrete time slots. Even when all 256 nodes in the entire warehouse are online simultaneously, it can completely eliminate packet collisions and packet loss caused by random channel contention, significantly improving the communication success rate in large-scale networking.

[0082] Step 3: All terminal devices that have not received a query command shall remain in a low-power sleep state.

[0083] Low-power sleep mode refers to a working mode in which the terminal device shuts down the LoRa RF transceiver unit and high-frequency clock source, retaining only the low-power timer running inside the MCU. For terminal devices with mismatched addresses or whose communication window has not yet been reached, their LoRa module remains powered off or in deep sleep, without attempting any carrier sensing or signal reception. This mechanism utilizes the TDMA (Time Division Multiple Access) concept, transforming passive listening into active wake-up, reducing the current consumption of the terminal device to the microampere level during non-working periods. Through this precise energy consumption control, compared to the traditional full-time listening mode, the average power consumption of the terminal device can be reduced by more than 70%, effectively extending the lifespan of monitoring equipment that relies on battery power or is deployed in areas with difficult power access, while also reducing electromagnetic background noise interference from non-target terminals to the wireless channel.

[0084] This application constructs a deterministic LoRa transparent polling system through the synergistic effect of the aforementioned technical features. The orderly command issuance from the host computer and the time-slot response from the terminal are tightly coupled, transforming the originally disordered, competitive random access into schedulable, ordered transmission. This not only solves the common problems of broadcast storms and air interface congestion in large-scale tobacco warehouse motor monitoring, ensuring that the data refresh latency across the entire warehouse remains stable at the second level, but also achieves optimal energy efficiency through a mandatory sleep strategy. This mechanism ensures that, even in complex electromagnetic environments, the active power, temperature, and humidity data of each motor can be reliably and with low latency transmitted back to the host computer, providing a high-quality data input foundation for subsequent refined classification of abnormal behavior and distributed impact domain assessment.

[0085] Example 4:

[0086] In another optional embodiment, the method includes a process for generating an anomaly classification information set, with the following specific steps:

[0087] Step 1: Establish sliding time series with a length of several polling periods for the motor active power, motor surface temperature, air outlet temperature and air outlet humidity respectively;

[0088] The sliding time series refers to a first-in, first-out (FIFO) data queue maintained by the host computer system for each monitoring point, used to store the raw parameter values ​​collected in the most recent polling cycles. The length of this sequence is set according to the inertial characteristics of environmental changes in the tobacco warehouse, typically set to 5 to 10 polling cycles to ensure that short-term fluctuations in parameters are captured while filtering out instantaneous noise. Specifically, whenever the host computer receives a new MODBUS response message from the terminal device via the LoRa gateway, it parses the current motor active power, motor surface temperature, outlet temperature, and outlet humidity, and sequentially pushes these values ​​to the end of the corresponding parameter's sliding time series, while removing the oldest data at the head of the queue to maintain a constant sequence length. For example, if the polling cycle is set to 30 seconds and the sequence length is set to 6, the sequence reflects the environmental evolution trajectory of this point over the past 3 minutes. Through this dynamic update mechanism, the system can grasp the historical trends of each physical quantity in real time, providing a continuous time-domain data foundation for subsequent trend analysis and avoiding misjudgments that may be caused by single-point sampling.

[0089] Step 2: Calculate the direction and amplitude of fluctuations in adjacent periods for each sliding time series to obtain the fluctuation trend vector for each parameter. Perform pairwise cross-comparison of the four fluctuation trend vectors to identify synchronous and asynchronous fluctuation modes.

[0090] The fluctuation direction refers to the change in the current period parameter value relative to the previous period parameter value, including three states: rising, falling, or remaining flat. The fluctuation amplitude refers to the absolute value of the difference between the current period parameter value and the previous period parameter value. The fluctuation trend vector is structured data composed of the fluctuation direction and fluctuation amplitude mentioned above, used to characterize the parameter's change characteristics on the time axis. Specifically, the host computer traverses each sliding time series and calculates the difference between data points at two adjacent time points. , where P t P is the parameter value for the current polling period. t-1 This is the parameter value from the previous polling cycle. If... ( (If the noise threshold is small), then the marking direction is upward, and the amplitude is... ;like If the value is high, it is marked as decreasing; otherwise, it is marked as remaining flat. After obtaining the fluctuation trend vectors of the four parameters, the system performs pairwise cross-comparison logic: comparing the motor active power vector with the motor surface temperature vector, comparing the outlet temperature vector with the outlet humidity vector, and comparing the motor surface temperature vector with the outlet temperature vector, etc. During this process, if two parameters show the same direction of change within the same time window (such as rising or falling simultaneously), it is identified as a synchronous fluctuation mode; if two parameters show opposite changes (such as one rising and the other falling) or the change sequence has a significant lag, it is identified as an asynchronous fluctuation mode. For example, when the motor active power is detected to be continuously rising and the motor surface temperature is rising synchronously, it is determined to be a power-temperature synchronous fluctuation mode; while when the motor surface temperature rises but the outlet temperature falls instead, it is determined to be a surface temperature-outlet temperature asynchronous fluctuation mode. This vector comparison-based mechanism can effectively extract the coupling relationship between multiple parameters and distinguish the linkage changes under normal operating conditions from the abnormal decoupling phenomenon under fault conditions.

[0091] Step 3: Based on synchronous and asynchronous fluctuation modes, classify abnormal states into heat dissipation obstruction type, condensation and water separation type, and system coupling failure type, and generate an abnormal classification information set for each abnormal point.

[0092] The anomaly classification information set is a structured data set containing anomaly location identifiers, anomaly type labels, descriptions of key fluctuation patterns that trigger the classification, and severity coefficients. Specifically, the system has a built-in classification rule engine that maps the identified fluctuation patterns to three predefined fault models: if the motor's active power and surface temperature show a synchronous upward trend, while the outlet temperature shows a asynchronous downward or stable trend, it is classified as a heat dissipation obstruction type, indicating that heat cannot be effectively dissipated; if the outlet temperature decreases and the outlet humidity increases synchronously and violently, while the motor power and surface temperature do not change significantly (i.e., they are asynchronous or static with other parameters), it is classified as a condensation type, indicating that over-cooling or refrigerant leakage has led to condensation; if the fluctuation trend vectors of three or more parameters simultaneously deviate from the historical baseline, and the correlation between them is disordered (i.e., synchronous and asynchronous modes are mixed and there is no clear physical logic), it is classified as a system coupling fault type, indicating that there are multiple concurrent problems within the equipment. For example, for a given location, if its fluctuation pattern exhibits asynchronous characteristics of simultaneous power increase and surface temperature increase while the outlet air temperature decreases, the system immediately generates a record labeled as "heat dissipation obstruction type," and records the fluctuation amplitude at this time as a severity coefficient, which is then incorporated into the anomaly classification information set. The generation of this information set marks the completion of the transformation from raw data to semantic fault diagnosis, providing clear input for subsequent impact domain assessment and significantly improving the interpretability and accuracy of fault identification.

[0093] This application achieves dynamic time-series modeling of motor operating status by constructing a sliding time series and extracting fluctuation trend vectors, overcoming the limitation of traditional single-threshold judgment in identifying progressive faults. By cross-comparing the fluctuation trend vectors of four key parameters pairwise, the system can accurately identify synchronous and asynchronous fluctuation modes. Utilizing the unique physical causal relationships in the tobacco storage environment (e.g., heat dissipation obstruction inevitably leads to a simultaneous increase in power and surface temperature, and condensation inevitably leads to a simultaneous decrease in outlet air temperature and a simultaneous increase in humidity), complex abnormal states are scientifically classified into heat dissipation obstruction type, condensation type, and system coupling fault type. This pattern recognition method based on multivariate time-series correlation not only effectively distinguishes between normal operating condition fluctuations and real hidden dangers, significantly reducing the false alarm rate, but also generates anomaly classification information sets containing rich semantic tags, providing reliable data support for subsequent distributed influence domain dynamic assessment and regional adaptive linkage control, achieving a technological leap from passive alarm to active diagnosis.

[0094] Example 5:

[0095] In one alternative embodiment, the method includes a process for determining heat dissipation obstruction anomalies.

[0096] Step 1: Extract the fluctuation trend vector of the motor's active power and the fluctuation trend vector of the motor's surface temperature, and calculate the trend consistency rate between the two. The trend consistency rate is defined as the proportion of the number of cycles in which the two fluctuate in the same direction to the total number of cycles.

[0097] The fluctuation trend vectors of the motor's active power and surface temperature are generated based on the sliding time series established in the above embodiments, respectively representing the energy input change trajectory and heat accumulation change trajectory of the motor within several consecutive polling cycles. The trend consistency rate is a key indicator for quantitatively evaluating the coupling strength of the motor's electro-thermal conversion. Its calculation logic involves traversing each cycle within the sliding window, counting the number of cycles in which active power and surface temperature rise or fall simultaneously, and dividing this number by the total number of cycles to obtain the ratio. Under normal operating conditions, an increase in motor load leads to a synchronous rise in temperature, but due to thermal inertia, there may be a brief phase difference between the two. In the initial stage of heat dissipation obstruction, because heat cannot be dissipated in time, the surface temperature response to power changes becomes more sensitive and synchronous. For example, if the sliding window length is set to 10 polling cycles, and if the fluctuation directions of the motor's active power and surface temperature are exactly the same (i.e., both positive or both negative growth) in 8 of these cycles, the calculated trend consistency rate is 0.8. By introducing trend consistency rate as the primary criterion, it is possible to effectively filter out random jumps in single parameters caused by instantaneous fluctuations in grid voltage or sensor noise, ensuring that the subsequent fine judgment process is only initiated when there is a strong correlation between power input and heat accumulation, thereby significantly reducing the false alarm rate.

[0098] Step 2: When the trend consistency rate exceeds the preset trend threshold, and the fluctuation trend vector of the air outlet temperature shows a continuous decrease, while the fluctuation amplitude of the motor surface temperature shows an accelerating increase, the current anomaly is classified as heat dissipation obstruction type.

[0099] The preset trend threshold is an empirical value derived from historical operating data of the tobacco warehouse air conditioning motor, used to define whether the electro-thermal coupling has reached an abnormal intensity; for example, it can be set to 0.75. This step constructs a triple constraint logic to accurately pinpoint heat dissipation obstruction faults: the first constraint is an excessive trend consistency rate, confirming a strong correlation between internal heat generation and increased motor surface temperature; the second constraint requires the outlet temperature fluctuation trend vector to show a continuous decrease, meaning that although the internal motor temperature is rising, the temperature of the blown-out cold air is decreasing or remaining low, intuitively reflecting that filter blockage, insufficient fan speed, or duct blockage prevents the effective removal of cold air from the motor cavity, forming the typical characteristic of internal heat and external cold; the third constraint requires the motor surface temperature fluctuation amplitude to show an accelerating increasing trend, that is, the second derivative of the temperature change is positive, indicating that the rate of heat accumulation at the motor casing is accelerating, and the fault is in a worsening stage. For example, if the trend consistency rate of a certain motor is detected to be 0.85 (exceeding the threshold of 0.75), and the outlet temperature drops from 16℃ to 14℃ and then to 13℃ (continuous decrease) in the last three cycles, while the surface temperature of the motor increases from 0.5℃ / min in the first cycle to 0.8℃ / min in the second cycle and then to 1.2℃ / min in the third cycle (accelerated increase), the system can determine that a heat dissipation obstruction anomaly has occurred at that point. This multi-dimensional joint judgment mechanism can strictly distinguish between normal load switching fluctuations and actual heat dissipation channel blockage, and is particularly suitable for identifying gradual heat dissipation failures caused by dust accumulation in tobacco warehouses.

[0100] This application constructs a rigorous system for identifying heat dissipation obstruction anomalies through the synergistic effect of the aforementioned technical features. By calculating the consistency rate between the trend of motor active power and surface temperature, non-coupled random interference is first eliminated at the source. Based on this, combining the dual characteristics of a continuous decrease in outlet temperature and an accelerated increase in surface temperature fluctuations, the system accurately captures the physical essence of normal heat generation but obstructed heat dissipation and uncontrolled heat accumulation. This progressive judgment logic solves the problem that traditional single-threshold alarms cannot distinguish between load fluctuations and actual heat dissipation faults, thus improving the accuracy and security of motor monitoring in tobacco warehouses.

[0101] Example 6:

[0102] In one embodiment, the method includes a process for determining the condensation and water separation type, with the following specific steps:

[0103] Step 1: Extract the fluctuation trend vectors of the air outlet temperature and humidity, and calculate the number of times the fluctuation direction reverses. The number of fluctuation direction reversals is defined as the number of times the temperature decrease segment and the humidity increase segment alternate within a time window.

[0104] The fluctuation trend vectors of the outlet temperature and humidity are generated based on the sliding time series established in the above embodiments, representing the dynamic changes in the thermodynamic and humidity states at the air conditioner outlet, respectively. The number of fluctuation direction reversals is a core indicator for quantifying the dynamic activity of condensation. Physically, it represents the frequency of complete cycles where the airflow temperature decreases followed by an increase in humidity within a preset time window (e.g., 30 minutes or 5 polling cycles). Specifically, the system first performs differential calculations on the outlet temperature sequence to identify the decreasing segment (i.e., the current cycle temperature is lower than the previous cycle), and simultaneously identifies the increasing segment of the outlet humidity sequence. Then, it counts the number of pairs of alternating occurrences of these two sequences on the time axis. For example, if the outlet temperature decreases for three consecutive cycles and the humidity increases for three consecutive cycles within a certain time window, it is recorded as a valid reversal event; if the temperature rises or the humidity decreases, the counting is interrupted. By calculating the number of reversals, the periodic condensation and defrosting behavior caused by overcooling on the evaporator surface can be keenly detected. This high-frequency alternation of cooling and humidification is often an early characteristic of refrigerant leakage or expansion valve failure, which is different from the smooth dehumidification process during normal refrigeration operation.

[0105] Step 2: When the number of times the fluctuation direction reverses exceeds the preset reversal threshold, and the time difference between the start time of the temperature drop and the start time of the humidity rise in the reversal event is less than the preset time difference threshold, and the fluctuation trend vectors of the motor active power and the motor surface temperature both show stable or random fluctuations, then the current anomaly is classified as condensation and water separation type.

[0106] This step ensures accuracy through multi-dimensional logical constraints, eliminating false alarms caused by non-condensation factors. The preset reversal threshold is an empirical value set based on the thermal inertia characteristics of the tobacco warehouse air conditioning unit, for example, 3 times / hour. Subsequent judgments are only triggered when the reversal frequency exceeds this threshold, filtering out occasional sensor noise. The introduction of the time difference threshold is based on thermodynamic causality: in a normal condensation process, the air must first be cooled below the dew point (the start of temperature drop), and then water vapor condenses, causing the humidity reading to rise (the start of humidity rise). Therefore, the start of temperature drop must be earlier than the start of humidity rise, and the time lag between the two should be within a physically reasonable range (e.g., less than two polling cycles, approximately 10 minutes). If the time difference is too large or the order is reversed, it indicates that the temperature and humidity changes are not directly coupled and will not be judged. Simultaneously, requiring the fluctuation trend vectors of motor active power and motor surface temperature to exhibit stable or random fluctuations is for the purpose of executing fault isolation logic: if the motor power increases significantly or the surface temperature fluctuates violently and synchronously, it may belong to the heat dissipation obstruction type or motor body failure mentioned in the above embodiment, rather than a simple condensation problem at the air handling end. Only when the motor operation is stable (power fluctuation amplitude is less than 5%, and the surface temperature change rate is close to zero), and the air outlet shows a violent temperature and humidity reversal, can it be confirmed as a condensation-type anomaly. For example, if the air outlet temperature at a certain point is monitored to drop from 18°C ​​to 14°C within 10 minutes, followed by a rise in humidity from 60% to 75%, and this process repeats 4 times, while the motor power remains stable at around 2.5kW and the surface temperature remains unchanged at 35°C, the system will classify it as a condensation-type anomaly. Thus, this step effectively distinguishes between equipment body failure and environmental control anomalies, significantly reducing the false alarm rate.

[0107] This application constructs a condensation diagnostic model based on a thermo-humidity coupling time sequence through the above steps. By calculating the number of reversals in the fluctuation direction of air outlet temperature and humidity, the activity level of condensation dynamics is accurately characterized. Utilizing the time sequence constraint that the temperature drop begins earlier than the humidity rise and the time difference is less than a threshold, the thermodynamic inevitability of cooling followed by moisture release is verified, effectively eliminating non-causal random fluctuation interference. Combined with the judgment of motor active power and surface temperature stability, abnormalities in the air handling process are successfully decoupled from motor failures. This judgment mechanism not only reduces the probability of misjudging normal operating fluctuations as faults but also accurately locates hidden dangers such as evaporator frosting and refrigerant leakage, providing a reliable decision-making basis for subsequent regional adaptive linkage control, thereby effectively curbing the risk of tobacco leaf mold caused by condensation and water release.

[0108] Example 7:

[0109] In another alternative embodiment, the method includes a system coupling failure determination process, specifically comprising the following steps:

[0110] Step 1: Calculate the first correlation between the motor's active power and the motor's surface temperature, the second correlation between the outlet temperature and the outlet humidity, and the third correlation between the motor's surface temperature and the outlet temperature.

[0111] Among them, correlation is an indicator used to quantify the degree of coordinated change between two different physical parameters during dynamic operation. The first correlation characterizes the coupling stability of the motor's electro-thermal conversion subsystem, reflecting whether the electrical energy input and thermal energy accumulation are matched; the second correlation characterizes the coupling stability of the air conditioning refrigeration system's heat-humidity exchange subsystem, reflecting whether the cooling process and dehumidification process are synchronized; the third correlation characterizes the coupling stability of the heat conduction subsystem that transfers heat from inside the equipment to the external environment, reflecting the degree of influence of motor heating on the outlet air temperature.

[0112] Specifically, the correlation is defined as the weighted sum of the square of the rate of consistency in the direction of fluctuation of two parameters within a sliding time window and the reciprocal of the difference in their fluctuation amplitudes. The rate of consistency in the direction of fluctuation can refer to the proportion of periods within a preset sliding time window (e.g., the last 5 polling periods) where the changing trends (rising, falling, or remaining flat) of the two parameters remain the same out of the total number of periods. This proportion ranges from [0,1], and squaring it aims to amplify the differences between completely synchronous or completely asynchronous characteristics. The difference in fluctuation amplitude can refer to the absolute value of the difference between the normalized changes of two parameters within the same time window. Taking its reciprocal means that the closer the changes in the amplitudes of the two parameters are, the larger this value is, indicating that they not only have the same direction but also match in magnitude.

[0113] For example, setting the sliding time window length to 5 polling cycles, for the calculation of the first correlation degree, if the motor's active power and surface temperature fluctuate in the same direction for 4 out of 5 cycles, the fluctuation direction consistency rate is 0.8, and its square is 0.64; if the difference in the average fluctuation amplitude after normalization is 0.1, its reciprocal is 10. The final first correlation degree value is obtained by weighting and summing using preset weighting coefficients (e.g., 0.6 for the direction term and 0.4 for the amplitude term). This definition method can simultaneously capture the trend synchronization and amplitude matching degree between parameters, effectively distinguishing between normal load fluctuations and abnormal system instability.

[0114] This step aims to construct a multi-dimensional system health fingerprint, upgrading the isolated monitoring of a single parameter to a comprehensive assessment of the interaction between multiple parameters, and providing a data foundation for identifying hidden system-level faults.

[0115] Step 2: The correlation is defined as the weighted sum of the square of the rate of consistency of the fluctuation direction of the two parameters within the sliding time window and the reciprocal of the difference in fluctuation amplitude;

[0116] This definition clarifies the mathematical logic of correlation calculation, ensuring the rigor of fault criteria. The square operation of the fluctuation direction consistency rate strengthens the signal characteristics of completely synchronized or completely divergent signals, making it impossible for noisy data that are occasionally in the same direction but are generally chaotic to obtain high scores; while the reciprocal operation of the fluctuation amplitude difference introduces the constraint of order-of-magnitude matching, preventing the occurrence of spurious correlation phenomena where the directions are the same but the amplitudes are vastly different.

[0117] Specifically, the historical normal range is derived from statistical analysis of correlation data accumulated during long-term fault-free operation of the equipment, typically expressed as an interval of mean plus or minus three standard deviations. The preset correlation threshold is a boundary value set based on the upper or lower limit of the historical normal range, used to define whether the correlation has deviated significantly. A certain number of polling cycles is a time-duration constraint used to filter transient interference; for example, it is set to three consecutive polling cycles. Only when the deviation continuously meets this duration requirement is it confirmed as a valid fault signal, thereby avoiding false alarms caused by instantaneous fluctuations in the power grid or brief sensor jumps.

[0118] For example, if historical data shows that the normal range for the first correlation degree is [0.75, 0.95], then the preset correlation degree threshold can be set to 0.70 (lower limit) and 1.00 (upper limit). When the real-time calculated first correlation degree is lower than 0.70 for three consecutive periods, it is determined that a deviation has occurred. This setting ensures that the system only triggers an alarm when it detects a continuous and significant multi-parameter decoupling phenomenon, significantly improving the accuracy of fault identification.

[0119] This step, by introducing persistent constraints and comparing statistical baselines, effectively eliminates the interference of random noise and ensures the reliability of the system coupling fault determination.

[0120] Step 3: When any two of the first correlation degree, the second correlation degree and the third correlation degree deviate from their respective historical normal ranges by more than the preset correlation degree threshold, and the deviation state is maintained for several consecutive polling cycles, the current anomaly is classified as a system coupling failure type.

[0121] The judgment logic employs a cross-validation mechanism that allows any two correlations to deviate simultaneously, aiming to capture complex, integrated failure modes. System-coupled failure anomalies are usually not caused by the failure of a single component, but rather by the combined effect of performance degradation in multiple subsystems (such as compressors, fans, and refrigerant circuits). A deviation in a single correlation may only indicate a local problem (such as a simple decrease in fan speed), but when two or more correlations deviate simultaneously, it strongly suggests a chain reaction or overall degradation within the system.

[0122] Specifically, a first correlation deviation may indicate decreased motor efficiency or poor heat dissipation; a second correlation deviation may indicate refrigerant leakage or heat exchanger frosting; and a third correlation deviation may indicate turbulent airflow or insulation failure. When any combination of these three factors (such as the first and second, the first and third, or the second and third) simultaneously exceeds the threshold and remains stable, the system is identified as a system coupling failure of the highest risk level.

[0123] For example, when the system detects a significant decrease in the first correlation (electricity-thermal) due to motor bearing jamming, and a significant decrease in the second correlation (thermal-humidity) due to asynchronous temperature and humidity control caused by compressor efficiency degradation, and this dual deviation persists for three polling cycles, the system will immediately mark the current state as a system-coupled fault type. This multi-correlation cross-validation strategy can improve the detection rate of system-level faults to an extremely high level, far exceeding traditional schemes that rely on single-parameter thresholds.

[0124] This step enables precise identification of complex failure modes, avoiding misjudging comprehensive system crashes as ordinary single anomalies, and providing solid evidence for subsequent advanced response strategies.

[0125] This application achieves real-time sensing of the entire coupling state of the electrical-thermal, thermal-humidity, and heat conduction links in the tobacco warehouse motor system by constructing a three-dimensional evaluation system with first, second, and third correlation degrees. By requiring any two correlation degrees to deviate simultaneously and for a certain period, and utilizing a cross-validation mechanism based on multi-dimensional features, it effectively overcomes the shortcomings of single-parameter monitoring, such as susceptibility to noise interference and difficulty in identifying complex faults. This combination of multi-correlation degree collaborative analysis and continuous criteria can accurately identify system coupled fault-type anomalies caused by the superposition of multiple factors such as compressor attenuation, bearing wear, and refrigerant insufficiency. This provides high-precision decision input for subsequent regional adaptive linkage control, thereby significantly improving the safety and reliability of the tobacco warehouse motor group operation.

[0126] Example 8:

[0127] One embodiment is a process for dynamic evaluation of distributed influence domains based on anomaly classification information set, which is provided in this application.

[0128] Step 1: Based on the coordinates of each abnormal point in the abnormal classification information set, retrieve the floor plan of the tobacco warehouse and identify other terminals in the vicinity of this point.

[0129] The tobacco warehouse floor plan refers to digital map data stored in the host computer system, which includes the physical coordinates of all air conditioning motor terminals, shelf distribution, and wall partition information within the warehouse. The adjacent area refers to a specific spatial range radiating outwards from the currently identified anomaly point. The initial radius of this range can be preset based on the warehouse floor height and equipment power, for example, set to 8 meters. The identification process involves calculating the Euclidean distance between the anomaly point coordinates and the coordinates of all other online terminals within the warehouse, filtering out the set of terminals falling within this initial radius. For example, when terminal AC-12 is marked as having a heat dissipation obstruction anomaly, the system automatically extracts its coordinates (X12, Y12) and traverses the entire warehouse terminal list, identifying terminals such as AC-09, AC-11, and AC-13, which are less than 8 meters from AC-12, as other terminals within the adjacent area. By combining static floor plan coordinates with dynamic anomaly labels, the spatial range affected by potential risks can be quickly identified, providing a physical basis for subsequent data correlation analysis.

[0130] Step 2: Extract the changes in air outlet humidity and motor surface temperature of each terminal in the vicinity within the most recent polling cycles, and count the number of terminals that show significant changes in the same direction.

[0131] The "recent polling cycles" can refer to a continuous time window tracing back from the current moment, such as the last 5 polling cycles (or the past 5 minutes if the polling interval is 1 minute). The change can refer to the difference between the parameter value in the current cycle and the value in the baseline cycle, or the first-order difference between adjacent cycles. A significant change in the same direction can mean that the parameter change trend of neighboring terminals is consistent with the parameter change trend of the anomaly point (e.g., both increasing or both decreasing), and the magnitude of the change exceeds a preset noise threshold (e.g., the absolute value of humidity change is greater than 2%RH, and the absolute value of temperature change is greater than 0.5℃). The statistical process involves iterating through all the identified neighboring terminals, comparing the time-series data of their air outlet humidity and motor surface temperature one by one. If at least one of these two parameters of a terminal shows a significant fluctuation in the same direction as the anomaly point, the count of that terminal is incremented by 1. For example, if the humidity at the air outlet of anomaly point AC-12 shows an upward trend, and the system detects that the humidity at the air outlet of neighboring terminal AC-13 also increased by 3% within the same time window, while the humidity of AC-09 and AC-11 remains stable or fluctuates randomly, then the number of terminals with significant changes in the same direction is counted as 1 (i.e., AC-13). This step, through data-level correlation verification, filters out devices that are spatially adjacent but have independent operating states, thereby accurately identifying the followers truly affected by the abnormal diffusion and effectively avoiding misjudgments caused by relying solely on distance.

[0132] Step 3: Based on the number of terminals, and combined with the airflow arrows marked on the tobacco warehouse floor plan and the location of warehouse partitions, determine the direction of abnormal diffusion and the boundary of the affected area;

[0133] The airflow arrow direction refers to the airflow vector pre-marked on the tobacco warehouse floor plan, reflecting the physical laws governing the air supply and return paths of the air conditioning system. Warehouse partition locations refer to physical barriers such as walls, fireproof roller shutters, or high-density shelving that obstruct airflow propagation. The determination process begins with a preliminary estimation of the influence radius based on the aforementioned number of terminals; the more terminals, the larger the preliminary radius. Then, the airflow arrow direction is used to anisotropically correct the preliminary radius: if a direction is downwind of the anomalous point (i.e., the airflow arrow points in that direction), the boundary of the influence domain for that direction expands outward (e.g., to 12 meters) to encompass any moisture or heat that may be dispersed by the airflow; if a direction is upwind, the boundary contracts inward (e.g., to 6 meters). Simultaneously, if a warehouse partition is encountered, regardless of the airflow direction, the boundary of the influence domain is cut off at the partition, as physical obstacles block the physical propagation of the anomalous system. For example, if statistics reveal three adjacent terminals exhibiting similar directional changes, and these terminals are primarily located on the east side of AC-12, while the plan shows the east side as upstream of the return air path without any obstruction, and the west side as blocked by a firewall, then the final determined boundary of the impact zone will extend eastward to 12 meters and terminate westward at the firewall. By introducing airflow dynamic constraints and physical obstruction limitations, the assessed boundary of the impact zone closely matches the actual risk spread path of the tobacco warehouse, significantly improving the accuracy of spatial early warning.

[0134] Step 4: Calculate the comprehensive risk index of all terminals within the boundary of the impact domain, and summarize the anomaly classification information and the impact domain boundary information to generate a distributed impact domain assessment information set.

[0135] The comprehensive risk index is a numerical indicator used to quantify the overall risk level within the influence domain. Its calculation relies on the real-time status data of each terminal within the influence domain and the attribute weights of the area. The calculation process involves traversing each terminal located within the defined influence domain boundary, extracting its current anomaly type, parameter deviation, and the tobacco stock value coefficient of its area, obtaining the risk score of each point through a weighted algorithm, and then aggregating all point scores (e.g., summing or taking the maximum value) to obtain the comprehensive risk index of the influence domain. The distributed influence domain assessment information set is a structured dataset that includes not only the calculated comprehensive risk index but also the generated anomaly classification labels (e.g., heat dissipation obstruction type), the defined influence domain boundary coordinate sequence, and the description of the diffusion direction. For example, the system calculates the comprehensive risk index of AC-12 and its influence domain (including AC-09, AC-11, and AC-13) to be 0.58, and packages this value with the heat dissipation obstruction type label and eastward diffusion boundary information to form a complete assessment record. This step elevates discrete point anomalies to a regional situation assessment. The output assessment information set directly reveals the spatial propagation intensity and potential hazard level of the anomalies, providing a precise spatial decision-making basis for subsequent implementation of hierarchical linkage control strategies.

[0136] This application achieves a leap from single-point anomaly monitoring to regional risk situation awareness through the synergistic effect of the above steps. By retrieving the floor plan of the tobacco warehouse and combining it with the coordinates of anomaly points, a spatial topological framework was constructed. Based on this, the humidity changes at the air outlets of adjacent terminals and the surface temperature changes of the motors were extracted, and the number of terminals showing significant changes in the same direction was counted. Data correlation was used to dynamically verify the authenticity and propagation of the risk, effectively eliminating spatially adjacent but logically unrelated interference items. Furthermore, the boundary of the impact zone was physically corrected by combining the direction of airflow arrows and the location of warehouse partitions, ensuring that the assessment results conform to the unique aerodynamic laws of tobacco warehouses, solving the problems of subjectivity and rigidity in impact range assessment in traditional methods. Finally, by calculating the comprehensive risk index within the impact zone and summarizing it to generate a distributed impact zone assessment information set, the abstract risk is transformed into a quantifiable index and a visualized boundary. This spatial-data-physical three-dimensional assessment mechanism not only improves the accuracy of anomaly diffusion identification but also provides data support for subsequent regional adaptive linkage control for different risk levels, ensuring that control commands can accurately cover the risk spread area, thereby effectively curbing the spread of fire or mold risks within the tobacco warehouse.

[0137] Example 9:

[0138] In another optional embodiment, the method includes a process for calculating a comprehensive risk index, specifically comprising the following steps:

[0139] Step 1: Based on the anomaly type of each anomaly point in the anomaly classification information set, retrieve the preset anomaly type weight table and assign a first weight value to this anomaly point.

[0140] The first weight value is a quantitative indicator representing the inherent hazard level of different failure modes to the tobacco storage environment. The preset anomaly type weight table is a mapping table pre-stored in the host computer system, established based on the probability and severity of secondary disasters caused by various anomalies in tobacco storage scenarios. Specifically, for heat dissipation obstruction anomalies, which are usually caused by filter blockage and develop relatively slowly, with the main risk being equipment overheating, a lower first weight value is assigned, such as 0.4. For condensation and water precipitation anomalies, which directly lead to a surge in humidity at the air outlet, easily causing mold and pests in tobacco leaves and posing a significant threat to tobacco quality, a higher first weight value is assigned, such as 0.7. For system coupling failure anomalies, which are often accompanied by multiple problems such as compressor efficiency decline and refrigerant leakage, easily inducing fires or large-scale shutdowns, the highest hazard level is assigned the highest first weight value, such as 1.0. Through table lookup operations, the system transforms abstract anomaly labels into numerical weights with clear business meanings. These weight values ​​directly reflect the inherent danger attributes of the fault itself, laying the foundation for subsequent risk assessment.

[0141] Step 2: Based on the boundary of the influence domain to which this point belongs, retrieve the preset regional weight table and assign a second weight value to this region;

[0142] The second weight value is a quantitative indicator characterizing the spatial spread range and intensity of the anomaly. A preset regional weight table defines the weight coefficients corresponding to the boundaries of different levels of influence domains. Specifically, when the influence domain determined by the distributed influence domain dynamic assessment is limited to the primary adjacent area around the anomaly location (e.g., within an 8-meter radius), it indicates a weak anomaly spread trend, and the second weight value is relatively small, for example, 0.3. When the influence domain extends to the secondary adjacent area or crosses warehouse partitions, it indicates that the anomaly has obvious thermal radiation conduction or humidity diffusion characteristics, and the second weight value is moderate, for example, 0.6. When the influence domain shows a cross-regional spread trend, involving multiple independent zones, it indicates a very high probability of risk spiraling out of control, and the second weight value is the largest, for example, 1.0. This step transforms the dynamic assessment results of the spatial dimension into numerical weights, enabling the risk index to keenly capture the evolution trend of the anomaly in physical space, achieving a quantitative transition from single-point monitoring to regional joint prevention.

[0143] Step 3: Obtain the stock value coefficient of tobacco storage areas within the boundary of the influence domain, normalized to the interval [0,1].

[0144] The stock value coefficient is a key parameter that maps technological risks into potential economic losses. The process of obtaining this coefficient involves: first, statistically analyzing the current tobacco inventory and corresponding grade value of all tobacco storage areas within the boundary of the impact domain; then, calculating the ratio of this total value to the maximum theoretical loss that a single incident in the warehouse could cause, or compressing it to a closed interval of [0,1] using a linear normalization function. For example, if the impact domain covers the core area storing high-grade aged tobacco leaves, its stock value is extremely high, and the calculated stock value coefficient is close to 1.0; if the impact domain only involves empty shelves or low-value turnover areas, the coefficient is close to 0.1. By introducing this coefficient, the risk assessment model no longer focuses solely on the operational status of the equipment itself, but also incorporates the economic value of the protected object, ensuring that even minor anomalies in high-value areas receive sufficient attention, reflecting a people-oriented and value-first management philosophy.

[0145] Step 4: Multiply the first weight value, the second weight value, and the stock value coefficient to obtain the comprehensive risk index.

[0146] The comprehensive risk index is the final risk quantification result obtained through three-dimensional coupling calculation. This step multiplies the first weight value representing the severity of the fault, the second weight value representing the breadth of diffusion, and the stock value coefficient representing economic value. This multiplicative logic means that the comprehensive risk index will only reach an extremely high level when the fault type is severe, the diffusion range is wide, and it involves high-value areas; conversely, if any dimension value is low, it will significantly lower the final index, thereby avoiding over-response to low-risk events. For example, when an anomaly point is determined to be a system-coupled fault type (first weight 1.0), its influence has spread to the secondary area (second weight 0.6), and it is located in the high-value tobacco storage area (stock value coefficient 0.9), the calculated comprehensive risk index is 0.54, which will directly trigger a high-level linkage control strategy. The resulting comprehensive risk index is not only a mathematical calculation result, but also a multi-dimensional decision-making benchmark that integrates equipment status, spatial propagation laws, and business value attributes, providing interpretable data support for the subsequent implementation of hierarchical linkage control.

[0147] This application utilizes the three-dimensional risk assessment model—hazard × breadth × value—constructed through the aforementioned steps to achieve a precise mapping of technical risks to business risks. The first weight value ensures the system's sensitivity to high-risk anomalies such as condensation-induced water separation and system coupling failures, avoiding response delays due to misjudgment. The second weight value dynamically captures the diffusion pattern of anomalies within the warehouse space, giving the risk assessment a spatiotemporal evolution characteristic. The stock value coefficient closely links cold, hard equipment data with the core asset value of the tobacco warehouse, making risk control strategies more aligned with actual production needs. The synergistic effect of these three factors ensures that the final comprehensive risk index objectively reflects the current safety situation and directly drives subsequent differentiated control commands. This solves the problems of abstract risk quantification and unclear handling guidance in traditional monitoring methods, improving the intelligence level and resource scheduling rationality of motor monitoring and control in tobacco warehouses.

[0148] Example 10:

[0149] In another optional embodiment, this application provides a process for generating a motor monitoring and control response information set. This method executes regional adaptive linkage control for different risk levels based on the distributed influence domain assessment information set generated in the aforementioned steps.

[0150] Step 1: Based on the comprehensive risk index of the distributed impact domain assessment information set, execute priority load reduction or shutdown control commands for the areas where abnormal points at the emergency response level are located;

[0151] The emergency response level is defined as a high-risk state based on the numerical range of the comprehensive risk index. Specifically, when the comprehensive risk index is within a preset high-risk threshold range (e.g., 0.6 to 0.8, or exceeding 0.8), the system determines that the location and its affected area are at the emergency response level. At this time, the host computer sends high-priority control commands to terminal devices within the target area via the LoRa gateway. For locations with a comprehensive risk index between 0.6 and 0.8, a zone isolation strategy is implemented, an emergency shutdown command is sent to forcibly cut off the motor power, and the remote restart function is locked to prevent accidental operation. For extreme cases where the comprehensive risk index exceeds 0.8 or is accompanied by a smoke alarm, the highest alert strategy for the entire warehouse is implemented, all polling is stopped, and audible and visual alarms are triggered. The load reduction control command specifically adjusts the motor operating mode to intermittent operation (e.g., running for five minutes, then stopping for two minutes) to reduce the heat load; the shutdown control command directly disconnects the solid-state relay to cut off the high-voltage circuit. Through this graded response mechanism, the spread of heat or moisture sources can be quickly blocked in the early stages of risk diffusion, preventing a single fault from evolving into a systemic accident.

[0152] Step 2: For areas not at the emergency response level, execute power-limited operation or ventilation adjustment instructions;

[0153] The non-emergency response level refers to a state where the comprehensive risk index is below the emergency threshold but abnormal characteristics still exist (e.g., an index between 0.3 and 0.6, or a specific type of mild anomaly such as heat dissipation obstruction). In this state, the system aims to eliminate potential hazards through gentle intervention, avoiding overreaction that could affect the overall stability of the warehouse environment. Specifically, the power-limiting operation command modifies the motor's target power limit, forcing the motor to operate under low load conditions, thereby reducing heat generation; the ventilation adjustment command controls the air conditioning fan to operate at a specific speed, accelerating local air circulation to remove accumulated heat or moisture. For example, when a comprehensive risk index of 0.45 is detected in a certain area and it is determined to be a condensation-type anomaly, the system automatically sends a load-reducing control command to that location and terminals within the adjacent primary influence zone, causing the motor to switch to intermittent operation mode. Simultaneously, the system marks the location in orange on the web-based host computer interface and displays a filter cleaning suggestion. This differentiated control strategy addresses potential equipment failures while maximizing the continuity of the tobacco storage environment.

[0154] Step 3: Record the sending time, target terminal address, and execution result status of each control command, and summarize them to form a motor monitoring and control response information set.

[0155] The motor monitoring and control response information set is a structured data set used to achieve closed-loop management and post-event traceability. Specifically, whenever the host computer issues a control command, the system immediately records the current timestamp as the sending time, parses and stores the unique identifier of the terminal device to which the command points (i.e., the target terminal address, usually corresponding to the MODBUS slave address or LoRa node ID). After the command is issued, the MCU of the terminal device executes the corresponding action (such as disconnecting the relay or activating the power limit) and transmits the success or failure status code back to the host computer via the LoRa network. This status is updated in real time to reflect the execution result. For example, if a stop command is sent to the terminal with address AC-12, the system records the sending time as 2023-10-13 14:30:05 and marks the status as executed after receiving the acknowledgment. Finally, all command records are summarized according to time sequence or regional distribution to form a complete log file. This not only ensures that every automated intervention is traceable, meeting the tobacco warehouse's requirements for operational traceability, but also provides real data support for subsequent optimization of risk thresholds and control strategies.

[0156] This application constructs a risk-driven adaptive control engine through the synergistic effect of the aforementioned technical features. The comprehensive risk index, as the core decision variable, is dynamically mapped to a differentiated control strategy library, achieving an automated closed loop from anomaly detection to precise response. The rapid shutdown at emergency response levels and the gentle adjustment at non-emergency levels work together to both contain the spread of disasters within seconds in the event of major risks and avoid overreacting to minor fluctuations, significantly reducing the frequency of manual intervention and maintenance costs. Simultaneously, the end-to-end instruction recording and status feedback mechanism ensures the traceability and reliability of control actions, solving the technical challenges of delayed control response, coarse-grained strategies, and lack of closed-loop verification in traditional tobacco warehouse monitoring.

[0157] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A distributed monitoring and control method for motors in a tobacco warehouse based on LoRa transparent polling, characterized in that, include: Obtain a multi-parameter information set for the tobacco warehouse, which includes motor active power, motor surface temperature, air outlet temperature, and air outlet humidity. Based on the multi-parameter information set of the tobacco warehouse, a refined classification of abnormal behaviors matching the tobacco storage environment is performed, generating an abnormal classification information set including abnormal tags such as heat dissipation obstruction type, condensation and water separation type, and system coupling failure type. Based on the aforementioned anomaly classification information set, combined with the tobacco warehouse floor plan and location data correlation, a distributed impact domain dynamic assessment is conducted to generate a distributed impact domain assessment information set containing a comprehensive risk index and impact domain boundary information. Based on the distributed impact domain assessment information set, regional adaptive linkage control is executed for different risk levels, generating a motor monitoring and control response information set containing control command sequences and execution status; The process of generating the distributed impact domain assessment information set includes: Based on the coordinates of each abnormal point in the abnormal classification information set, retrieve the tobacco warehouse floor plan and identify other terminals in the vicinity of this point. Extract the changes in air outlet humidity and motor surface temperature of each terminal in the neighboring area within the most recent polling cycles, and count the number of terminals that show significant changes in the same direction. Based on the number of terminals, and combined with the airflow arrow directions marked on the tobacco warehouse floor plan and the warehouse partition locations, the abnormal diffusion direction and the boundary of the affected area are determined. Calculate the comprehensive risk index of all terminals within the boundary of the influence domain, and summarize the anomaly classification information and the influence domain boundary information to generate the distributed influence domain assessment information set; The calculation process of the comprehensive risk index includes: Based on the anomaly type of each anomaly point in the anomaly classification information set, a preset anomaly type weight table is retrieved, and a first weight value is assigned to this anomaly point. Based on the boundary of the influence domain to which this point belongs, retrieve the preset regional weight table and assign a second weight value to this region; Obtain the stock value coefficient of tobacco storage area within the boundary of the influence domain, normalized to the interval [0,1]. The comprehensive risk index is obtained by multiplying the first weight value, the second weight value, and the stock value coefficient.

2. The method according to claim 1, characterized in that, The process of acquiring the multi-parameter information set of the tobacco warehouse includes: The SPI interface built into the MCU core control module of each terminal device is used to communicate with the BL0942 power metering chip to read the active power of the motor in real time. Simultaneously, through a universal input / output interface, data is read from the motor surface temperature and humidity sensor, the air outlet temperature and humidity sensor, and the smoke sensor, respectively, to collect the motor surface temperature, the air outlet temperature, and the air outlet humidity; The MCU core control module assembles the raw data into a response message according to the MODBUS protocol format, transmits it to the LoRa wireless transparent transmission module through the optically isolated serial port unit, and then transmits it back to the host computer via the LoRa gateway and Ethernet, thus summarizing and forming a multi-parameter information set covering the entire tobacco warehouse.

3. The method according to claim 2, characterized in that, The LoRa wireless transparent transmission module uses a polling mechanism based on time slot allocation when performing data backhaul: The host computer sends query commands to each terminal device in sequence at a preset polling cycle. After receiving the command corresponding to its own address, each terminal device only starts the LoRa module to respond and upload within the allocated time slot. All terminal devices that do not receive query commands remain in a low-power sleep state.

4. The method according to claim 2, characterized in that, The process of generating the anomaly classification information set includes: Establish sliding time series with a length of several polling periods for the active power of the motor, the surface temperature of the motor, the air outlet temperature, and the air outlet humidity. For each sliding time series, the fluctuation direction and fluctuation amplitude of adjacent periods are calculated to obtain the fluctuation trend vector of each parameter. The four fluctuation trend vectors are cross-compared pairwise to identify synchronous fluctuation patterns and asynchronous fluctuation patterns. Based on the synchronous fluctuation mode and the asynchronous fluctuation mode, the abnormal state is classified into the heat dissipation obstruction type, the condensation and water separation type and the system coupling failure type, and an abnormal classification information set is generated for each abnormal point.

5. The method according to claim 4, characterized in that, The process for determining the type of heat dissipation obstruction includes: Extract the fluctuation trend vector of the active power of the motor and the fluctuation trend vector of the surface temperature of the motor, and calculate the trend consistency rate of the two. The trend consistency rate is defined as the proportion of the number of cycles in which the two fluctuate in the same direction to the total number of cycles. When the trend consistency rate exceeds the preset trend threshold, and at the same time the fluctuation trend vector of the air outlet temperature shows a continuous decrease, while the fluctuation amplitude of the motor surface temperature shows an accelerating increase, the current anomaly is classified as the heat dissipation obstruction type.

6. The method according to claim 4, characterized in that, The process for determining the condensation and water separation type includes: Extract the fluctuation trend vector of the air outlet temperature and the fluctuation trend vector of the air outlet humidity, and calculate the number of times the fluctuation direction reverses. The number of fluctuation direction reversals is defined as the number of times the temperature drop segment and the humidity rise segment alternate within a time window. When the number of reversals in the direction of fluctuation exceeds a preset reversal threshold, and the time difference between the start time of the temperature drop and the start time of the humidity rise in the reversal event is less than a preset time difference threshold, and the fluctuation trend vectors of the motor's active power and the motor's surface temperature both show stable or random fluctuations, then the current anomaly is classified as the condensation and water separation type.

7. The method according to claim 4, characterized in that, The process for determining the system coupling fault type includes: Calculate the first correlation between the active power of the motor and the surface temperature of the motor, the second correlation between the outlet temperature and the outlet humidity, and the third correlation between the surface temperature of the motor and the outlet temperature; The correlation is defined as the weighted sum of the square of the rate of consistency of the fluctuation direction of the two parameters within the sliding time window and the reciprocal of the difference in fluctuation amplitude. When any two of the first correlation degree, the second correlation degree, and the third correlation degree simultaneously deviate from their respective historical normal ranges by more than a preset correlation degree threshold, and the deviation state is maintained continuously for several polling cycles, the current anomaly is classified as the system coupling failure type.

8. The method according to claim 1, characterized in that, The process of generating the motor monitoring and control response information set includes: Based on the comprehensive risk index in the distributed impact domain assessment information set, priority load reduction or shutdown control commands are executed for the areas where abnormal points at the emergency response level are located. For areas not at the emergency response level, implement power-limited operation or ventilation adjustment instructions; The sending time, target terminal address, and execution result status of each control command are recorded and summarized to form the motor monitoring and control response information set.

Images