An intelligent control system

Abstract

The application relates to the control field and discloses an intelligent control system for improving the accuracy, reliability and adaptability of remote automatic equipment start-stop control, which can be adapted to various scenes such as televisions and washing machines that need intelligent control and is used for realizing intelligent control of equipment and comprises a collection module, an identification module, a processing module and a control module. The collection module is responsible for comprehensively collecting optocoupler-isolated remote control signals, external state detection signals and real-time load currents and constructing a multi-dimensional equipment state data set. The identification module accurately determines the chaotic characteristics of the system by using phase space reconstruction technology and Lyapunov index calculation. When the chaotic characteristics are detected, the processing module generates fault prediction state data based on a prediction model and triggers a protection mechanism under abnormal conditions. The control module transmits predictive protection control instructions to the controlled equipment. The system takes comprehensive control as the core and provides an intelligent and predictive protection scheme for the equipment.

Description

Technical Field

[0001] This invention relates to the field of control, and more particularly to an intelligent control system. Background Technology

[0002] With the rapid development of IoT technology and the continuous improvement of industrial automation, the start-stop control and management of remote automated equipment has become a core requirement in modern industrial production, energy management, and smart buildings. Traditional equipment start-stop control methods mainly rely on localized manual operation or simple timed control strategies, which suffer from slow response speed, delayed fault detection, and high maintenance costs. Especially in complex industrial environments, equipment operating status is affected by various factors, such as load fluctuations, environmental interference, and mechanical wear. These factors may cause the equipment operating status to exhibit nonlinear and chaotic characteristics, leading to unpredictable faults or shutdowns, seriously affecting production efficiency and system stability.

[0003] In recent years, the introduction of IoT technology has provided new solutions for remote monitoring and management of equipment. Through IoT management systems, equipment status data can be uploaded to the cloud in real time, enabling remote monitoring, fault diagnosis, and predictive maintenance. However, existing technologies still have the following shortcomings:

[0004] Traditional methods collect only a limited number of equipment status parameters, which makes it difficult to fully reflect the equipment's operating status, especially in terms of monitoring nonlinear behaviors such as chaotic characteristics.

[0005] Existing technologies are mostly based on threshold alarms or simple trend analysis, which cannot accurately predict the future state of equipment. Especially in chaotic states, the accuracy and timeliness of fault prediction are difficult to guarantee.

[0006] Traditional protection measures are mostly passively triggered and cannot actively intervene before a fault occurs, resulting in a high risk of equipment damage.

[0007] The control parameters of existing technologies are mostly fixed and cannot be dynamically adjusted according to the equipment operating environment or historical fault data, resulting in a high false alarm rate or missed alarm rate under complex operating conditions.

[0008] Therefore, we propose an intelligent control system to solve the above problems. Summary of the Invention

[0009] This invention provides an intelligent control system for improving the accuracy, reliability, and adaptability of start-stop control for remote automated equipment.

[0010] The first aspect of this invention provides an intelligent control system, comprising: a data acquisition module for acquiring remote control signals, acquiring external state detection signals via pins, and sampling load current in real time; an identification module for generating a multi-dimensional device state dataset based on the remote control signals, external state detection signals, and sampled load current, performing phase space reconstruction mathematical operations on the multi-dimensional device state dataset, calculating the Lyapunov exponent of the load current sequence through the floating-point unit built into the main control chip, and outputting a chaotic characteristic judgment flag; a processing module for predicting future operating states based on the reconstructed phase space and generating fault prediction state data when the chaotic characteristic judgment flag indicates chaotic characteristics; and a control module for triggering a channel switching instruction set, generating a predictive protection control instruction, and transmitting the predictive protection control instruction to the controlled device if the fault prediction state data indicates an abnormality.

[0011] Optionally, in a first implementation of the first aspect of the present invention, the multi-dimensional device state dataset is set as follows: ,but:

[0012] ;

[0013] in, The TTL level vector of the remote control signal after optical coupler conversion at time t; It is a remote function;

[0014] Let be the state signal vector synchronously read at time t, where It is a fault signal. It is a low-order signal. It is a high-level signal. It's a door lock signal;

[0015] Let be the state function, calculated as The range of values ​​is {0,1,…,15}; The shunt resistor voltage value acquired by the ADC channel at time t;

[0016] Let be a function of current, calculated as ,in This is the ADC full-scale voltage, corresponding to the maximum load current. ,and , It is the shunt resistor value.

[0017] Optionally, in a second implementation of the first aspect of the present invention, the method includes: extracting a load current time sequence from the multi-dimensional device state dataset; sampling the current sequence at equal intervals using a preset time delay parameter to construct a three-dimensional phase space vector set and generate a reconstructed phase space point set; locating nearest neighbor vector pairs in the reconstructed phase space point set, calculating the ratio of the initial distance to the evolved distance, and generating a distance ratio sequence; performing logarithmic operations on the distance ratio sequence using the floating-point unit of the main control chip, accumulating the results using a sliding window, and outputting a Lyapunov exponent value; comparing the Lyapunov exponent value with a preset positive and negative threshold: if it is greater than the positive threshold, outputting a chaos flag; if it is less than the negative threshold, outputting a stability flag; otherwise, maintaining the previous state flag and generating a chaos characteristic judgment flag.

[0018] Optionally, in a third implementation of the first aspect of the present invention, the method includes: selecting the nearest neighbor set of the current state point from the reconstructed phase space point set; combining the high-level state signal and the door lock state signal to fit a multidimensional evolution model to predict the contact impedance change trend and generate a contact state prediction value; obtaining a dynamic threshold parameter; comparing the contact state prediction value with the dynamic threshold to generate fault prediction state data; generating a channel switching instruction set based on the fault prediction state data; reading the current value of the fault counter; dynamically adjusting the parameter weights of the prediction model according to the fault count to generate adaptive prediction parameters; combining the channel switching instruction set with the adaptive prediction parameters; adding a device identifier to generate a predictive protection control instruction.

[0019] Optionally, in a fourth implementation of the first aspect of the present invention, the contact state prediction value is set to... : ;in, Let t be the state variable at the current time. Let be the state variables at historical time t-τ. For time delay; The state variables are for an earlier historical moment t-2τ.

[0020] Optionally, in the fifth implementation of the first aspect of the present invention, the method includes: extracting the control field and cooling field from the predictive protection control instruction and merging them into an 8-bit original control word; obtaining the lower 4 bits of the current system timestamp as a time identifier, performing an XOR operation between the higher 4 bits of the 8-bit original control word and the time identifier to generate a dynamic check code; writing the dynamic check code into the lower 4 bits, retaining the lower 4 bits of the 8-bit original control word, and combining them to generate a 1-byte simplified status code; utilizing multi-protocol adaptive transmission, when no confirmation is received from the IoT platform after 3 consecutive transmissions, triggering the status code backup storage to EEPROM to generate an offline storage record.

[0021] Optionally, in the sixth implementation of the first aspect of the present invention, an update module is further included, used to receive control parameters issued by the IoT management system, update the Lyapunov exponent threshold, and generate an adaptive parameter set: receiving Modbus RTU format parameter packets through an RS485 interface, parsing UDP parameter packets through an Ethernet interface, or obtaining transparent parameter data through a 4G module, extracting the original hexadecimal parameter stream, and generating a parameter data packet; separating a verification flag from the parameter data packet, triggering a CRC verification module to verify data integrity, and generating a threshold parameter group that has passed verification; reading the current value of the fault counter, dynamically scaling the threshold parameter group according to the fault count level, and generating an environment-corrected threshold parameter; combining the environment-corrected threshold parameter with the device operating mode flag, adding a version number and a timestamp identifier, and generating an adaptive parameter set; writing the adaptive parameter set to the starting address 0x20002000 of the SRAM, sending a parameter reception confirmation signal to the IoT management system, and generating a parameter update completion flag.

[0022] Optionally, in the seventh implementation of the first aspect of the present invention, the verification flag includes: a high 8-bit Lyapunov positive threshold, a low 8-bit negative threshold, and 2 bits; the dynamic scaling of the threshold parameter group according to the fault count level includes: when the fault count > 10, the threshold is amplified by 120%; when the fault count < 5, the threshold is reduced by 80%.

[0023] Beneficial effects: By synchronously acquiring data through PA08-PA11, PB22-PB28 and ADC channels of the main control chip, covering control commands, equipment status and electrical parameters, it breaks through the limitations of single parameter monitoring in traditional methods, realizes joint analysis of control commands, mechanical status and electrical characteristics, improves the accuracy of fault diagnosis, and eliminates misjudgment caused by asynchronous sampling through timing alignment and data fusion.

[0024] Traditional methods can only detect linear faults, while this invention can identify potential faults in chaotic states, bringing the fault detection time forward to the nascent stage. Through predictive maintenance, it reduces sudden downtime caused by chaotic behavior and improves production continuity.

[0025] Dynamic threshold adjustment enables the system to maintain high-precision prediction under different fault frequencies and operating environments, while remote parameter updates reduce on-site maintenance needs and improve the system's intelligence level. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of one embodiment of the intelligent control system in this invention;

[0027] Figure 2 This is a schematic diagram of one embodiment of the intelligent control method in this invention;

[0028] Figure 3 This is a schematic diagram of another embodiment of the intelligent control system in this invention. Detailed Implementation

[0029] This invention provides an intelligent control system for improving the accuracy, reliability, and adaptability of start-stop control for remote automated equipment. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0030] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent control system in this invention includes:

[0031] 101. Acquisition module, used for multi-source signal acquisition and fusion: acquires remote control signals with optical isolation through PA08-PA11 pins of the main control chip ATSAMC21J17A-Z, acquires external status detection signals through PB22, PB23, PA27, PA28 pins, and samples load current in real time using the ADC channel;

[0032] It is understood that the executing entity of this invention can be an intelligent control device, a terminal, or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as an example.

[0033] It should be noted that in industrial control systems, the main control chip ATSAMC21J17A-Z is used to acquire and fuse multi-source signals, generating a multi-dimensional dataset that includes remote control commands, equipment status, and load current, providing input for subsequent chaos analysis.

[0034] Signal acquisition hardware configuration: optocoupler-isolated remote control signals (4-channel digital signals); Pins: PA08-PA11 (configured as digital input); Signal source: start / stop commands from the IoT management system (high level = start, low level = stop). Example data: PA08: 3.3V (start command); PA09: 0V (stop command); PA10: 3.3V (fault reset); PA11: 0V (spare); Isolation protection: uses HCPL-3700 optocouplers, with an isolation voltage of up to 5kV, suppressing electromagnetic interference.

[0035] External status detection signals (4-channel analog / digital signals), pins: PB22 (temperature sensor), PB23 (vibration sensor), PA27 (voltage fluctuation), PA28 (humidity sensor);

[0036] Example data: PB22: 1.2V (corresponding temperature 85℃); PB23: 2.8V (vibration amplitude 0.5mm); PA27: 3.0V (voltage fluctuation ±10%); PA28: 0.8V (humidity 60%RH);

[0037] Load current sampling (ADC channel), configuration: ADC channel 0 (PA02 pin) is connected to a current transformer (100:1 ratio). Sampling parameters: Sampling rate: 1kHz (real-time capture of transient current); Range: 0–5A (corresponding to ADC output 0–3.3V); Example data: Steady-state current: 2.1A ( V); Inrush current: 4.7A ( V (lasts for 50ms);

[0038] Data preprocessing and fusion, data standardization: Analog signals are uniformly standardized using Z-score (formula: Eliminate dimensional differences. Temperature signal (85℃) → mean. ℃, standard deviation → Digital signals are directly converted to Boolean values ​​(0 / 1).

[0039] Noise suppression: Kalman filtering is applied to load current signals to reduce sampling noise (process noise covariance). Observation noise After filtering, the peak surge current was corrected from 4.7A to 4.5A (error <5%).

[0040] Spatiotemporal alignment: Timestamp synchronization: All signals are aligned based on the ADC sampling time (vibration signal delay of 2ms, compensated by interpolation). Data frame structure: A 7D dataset (4 control signals + 4 status signals + 1 current) is generated every 10ms.

[0041] Generate a multi-dimensional device status dataset. A data example (single frame data) is shown in Table 1.

[0042] Table 1. Multi-dimensional Device Status Dataset (Single Frame Data)

[0043] signal type parameter Sample value Standardized value Remote control (PA08) Startup command 3.3V 1 (True) Temperature (PB22) Sensor voltage 1.2V 1.0 (Z-score) Vibration (PB23) amplitude 2.8V 1.4 (Z-score) Load current (ADC0) effective value 2.1A 0.8 (Z-score)

[0044] Data set uses: 4-channel control signals → to determine the validity of operation commands; temperature / vibration / voltage → to assess the health status of equipment; load current → to detect overload or open circuit risks.

[0045] 102. Identification module, used for chaotic feature mathematical operations: generates a multi-dimensional device state dataset, performs phase space reconstruction mathematical operations on the multi-dimensional device state dataset, calculates the Lyapunov exponent of the load current sequence through the built-in floating-point unit of the main control chip, and outputs the chaotic characteristic judgment flag.

[0046] It should be noted that when a certain industrial equipment experiences abnormal fluctuations, the main control chip ATSAMC21J17A-Z is used to perform chaotic characteristic analysis on the load current sequence to determine whether there is a risk of chaotic oscillation.

[0047] Data preprocessing, input data: from the multi-dimensional dataset obtained in step 101, the load current sequence is extracted as the analysis object (sampling rate 1kHz, duration 10s, total 10,000 points).

[0048] Example data characteristics: Steady-state current: 4.8A (normal range ±0.2A); Abnormal fluctuations: intermittent oscillations occur (amplitude suddenly increases to 7.2A, lasts for 50ms and then recovers).

[0049] Phase space reconstruction, parameter determination: Time delay τ: calculated by the autocorrelation function method to reduce the current sequence to its decay state. In approximately 0.367 seconds, we obtained... ms (i.e., 4 sampling points). Embedding dimension m: determined using the pseudo nearest neighbor (FNN) method. (Satisfies the requirements of Takens' theorem).

[0050] Reconstruction process: converting a one-dimensional current sequence Convert to a three-dimensional phase space vector: Generate a reconstructed phase space matrix (9992x3 dimensions), with each row representing a phase point.

[0051] Lyapunov exponent calculation, algorithm selection: Wolf algorithm (suitable for real-time calculation in embedded systems) is adopted.

[0052] Calculation process: Search for adjacent phase points in the reconstructed phase space (initial distance) A). Track the evolution of adjacent orbits and calculate the distance change rate within 1000 steps. Logarithmic operations and mean calculations are accelerated using the main control chip's built-in FPU (floating-point unit). Output: Lyapunov exponent. (Accuracy ±0.05).

[0053] Chaotic property determination, determination logic: if If so, the system exhibits chaotic characteristics. Flag generation: In this example... → Output the chaos flag as "1" (1-byte binary flag). Compare to the threshold: When the state is stable, output "0".

[0054] 103. Processing module, used for predictive control instruction generation: When the chaotic characteristic determination flag indicates chaotic characteristics: predict the future working state based on the reconstructed phase space and generate fault prediction state data; if the fault prediction state data indicates an abnormality, trigger the channel switching instruction set through the PB12-PB15 pins to generate predictive protection control instructions.

[0055] It should be noted that a certain industrial equipment triggered a chaotic characteristic determination flag during operation. The system predicts future states based on the reconstructed phase space and performs protection operations.

[0056] Reconstructing the phase space prediction state, input data: multi-dimensional dataset (generated in step 101): including load current sequence (1kHz sampling, 10-second data), temperature (85℃), vibration amplitude (0.5mm), etc. Reconstructing phase space parameters (generated in step 102): time delay. ms, embedded dimension .

[0057] Prediction method: In three-dimensional phase space, using the current phase point Starting from this point, search for the nearest neighbor phase trajectory. Track the evolution trend of adjacent trajectories and predict the load current change within the next 5ms (steady-state current 4.8A → predicted peak 7.2A, lasting 50ms).

[0058] The fault prediction status data is generated as shown in Table 2:

[0059] Table 2 Fault Prediction Status Data

[0060] Predicted features status code value meaning Current surge >40% 0x01 Overload risk Current oscillation frequency >100Hz 0x02 Mechanical resonance Temperature + Current Co-exceeding Limits 0x03 Insulation aging

[0061] In this example, the output status code is 0x01 (Current prediction peak 7.2A, exceeding the rated value by 44%).

[0062] Switching protection operation, trigger condition: fault prediction status data ≠ 0 (status code 0x01 in this example indicates an anomaly). Hardware action: Pin operation: Output 4 timing signals through pins PB12-PB15: (Power outage); (power ups); (Mechanical latch release); (Bypass capacitor engaged); Switching effect: The load is switched to the redundant circuit, the current drops to a safe value of 3.0A, and the response time is <10ms.

[0063] Protection instruction feedback: Generate an 8-bit predictive protection control instruction word (10110010), where: the high 4 bits: status code (0x01→"0001"); the low 4 bits: action type (switch→"0010").

[0064] 104. Control module for IoT communication optimization transmission: The predictive protection control command is processed into a 1-byte simplified status code and transmitted to the IoT management system through at least one of the following communication interfaces: 4G wireless module controlled by PA00-PB09 pins, Ethernet interface controlled by PB00-PB03 pins, and RS485 bus of PA24-PA25 pins.

[0065] It should be noted that after a certain industrial equipment triggers the switching protection, the system needs to efficiently transmit the protection command to the IoT management platform.

[0066] Protection command optimization, input command: predictive protection control command word (8-bit binary, 10110010); high 4 bits: fault status code (overload risk 0x01→0001); low 4 bits: action type (switching→0010).

[0067] 1-byte simplified processing: Compressed into a single-byte (8-bit) status code, directly mapped to the instruction word (10110010→0xB2). Advantages: Data volume reduced to 1 / 8 (originally requiring 8 bytes to describe actions and status, now only 1 byte); Strong compatibility: 1 byte can cover 256 fault-action combinations, meeting the needs of industrial scenarios.

[0068] Multi-channel communication interface transmission, 4G wireless transmission (PA00-PB09 pin control): Module model: 4GCat.1 module (supports TCP / IP and MQTT protocols); Data packet structure: Destination address (IoT platform IP) + status code 0xB2 + timestamp; Total length: 20 bytes (including protocol header). Example transmission latency: Average time from instruction generation to platform reception is 85ms (measured value).

[0069] Ethernet interface (PB00-PB03 pin control): Protocol: ModbusTCP, directly transmits status code 0xB2 to the local server; Bandwidth usage: only 1.5% of the 100Mbps Ethernet bandwidth is used per transmission.

[0070] RS485 bus (PA24-PA25 pins): Topology: Bus network connecting multiple devices (1 master and multiple slaves); Transmission efficiency: At a baud rate of 115200bps, status code transmission takes only 0.07ms.

[0071] Communication reliability assurance mechanism, multi-channel hot backup: 4G transmission is given priority, and Ethernet or RS485 is automatically switched when the signal strength is <-90dBm.

[0072] Data integrity verification: Add CRC-8 check bits (1 byte) to ensure that the status code is transmitted correctly (0xB2 check code is 0x1C).

[0073] Anti-interference design: RS485 uses differential signal transmission to suppress common-mode interference (measured bit error rate in industrial environment <10%). -6 Performance indicators are shown in Table 3:

[0074] Table 3 Performance Indicators

[0075] Communication methods Transmission delay Applicable Scenarios 4G wireless <100ms Remote cross-regional monitoring Ethernet <2ms Factory local high-speed network RS485 <0.1ms Densely populated areas (<1km)

[0076] In this embodiment of the invention, remote control signals and external status detection signals isolated by optocouplers are acquired through specific pins of the main control chip, and the load current is sampled in real time using the ADC channel to generate a multi-dimensional dataset containing remote control commands, device status, and load current. This dataset can accurately reflect the device's operating status and provide rich and accurate data support for subsequent analysis, effectively avoiding misjudgments and omissions caused by monitoring a single signal. The use of HCPL-3700 optocouplers to isolate the remote control signals can effectively suppress electromagnetic interference and improve the stability and reliability of signal acquisition. Meanwhile, pins and sensors are rationally configured for different types of signals to ensure accurate acquisition of various signals; Kalman filtering is used to reduce sampling noise of load current signals, and spatiotemporal alignment is achieved through interpolation compensation, generating a 7D dataset frame every 10ms, improving data quality and usability, and providing a high-quality data foundation for subsequent analysis; the load current sequence is reconstructed in phase space, and the time delay and embedding dimension are determined by the autocorrelation function method and the pseudo nearest neighbor method, converting the one-dimensional current sequence into a three-dimensional phase space vector and generating a reconstructed phase space matrix; the main control chip has a built-in floating-point unit to accelerate logarithmic operations and mean calculation, improving the Lyapunov exponent calculation speed, ensuring the real-time nature of chaotic characteristic analysis, and enabling timely detection of potential problems in equipment operation, thus gaining time for timely and effective control measures; when the fault prediction status data indicates anomalies, a channel switching instruction set is triggered through a specific pin to generate predictive protection control instructions. The load is switched to redundant circuits, the current drops to a safe level, and the response time is less than 10ms, enabling rapid disconnection of the fault source, protecting equipment safety, reducing equipment damage and downtime, and minimizing production losses. A multi-channel hot backup mechanism is employed, prioritizing 4G transmission and automatically switching to Ethernet or RS485 when signal strength is low to ensure communication continuity. A CRC-8 checksum is added to ensure data integrity, and RS485 uses differential signal transmission to suppress common-mode interference, effectively guaranteeing communication quality. This allows the IoT management system to obtain equipment status information in a timely and accurate manner, enabling remote monitoring and management.

[0077] Please see Figure 3 Another embodiment of the intelligent control system in this invention includes:

[0078] 101. Acquisition module, used for multi-source signal acquisition and fusion: acquires remote control signals with optical isolation through PA08-PA11 pins of the main control chip ATSAMC21J17A-Z, acquires external status detection signals through PB22, PB23, PA27, PA28 pins, and samples load current in real time using the ADC channel;

[0079] 102. Identification module, used for chaotic feature mathematical operations: generates a multi-dimensional device state dataset, performs phase space reconstruction mathematical operations on the multi-dimensional device state dataset, calculates the Lyapunov exponent of the load current sequence through the built-in floating-point unit of the main control chip, and outputs the chaotic characteristic judgment flag.

[0080] Specifically, the system features include: adaptive optocoupler signal conversion: converting the remote control signals of pins PA08-PA11 to TTL level using a COSMO1010 optocoupler to generate a standard remote command set; synchronous acquisition of status detection signals: synchronously reading the fault signal status of pin PB22, the low-order signal status of pin PB23, the high-order signal status of pin PA27, and the door lock signal status of pin PA28 within a 10ms sampling period to generate a four-dimensional status vector; and dynamic quantization of load current: acquiring the load current of the ULN2803 drive circuit at a sampling rate of 500Hz via an ADC channel. The shunt resistor voltage is converted in real time based on a pre-stored resistance-current mapping table to generate a load current timing sequence. Timestamp alignment and fusion: Millisecond-level timestamps are added to the standard remote instruction set, four-dimensional state vector, and load current timing sequence, and timing alignment is performed through a circular buffer to generate a time-stamped fused dataset. Data dimension compression processing: The time-stamped fused dataset is processed according to the following rules: the remote instruction set is mapped to a 2-bit binary code, the four-dimensional state vector is compressed to a 4-bit status word, and the load current value is quantized to an 8-bit integer, generating a multi-dimensional device status dataset. And store it in the SRAM buffer.

[0081] ;

[0082] in, At time t, the remote control signal TTL level vector after optocoupler conversion, each The vector ∈{0,1} corresponds to the signals on pins PA08-PA11. This vector is generated when the remote control signal is converted to TTL level using a COSMO1010 optocoupler and represents the raw input of the standard remote command set.

[0083] For remote functions, The mapping is to a 2-bit binary code with a value range of {0, 1, 2, 3}. This function is implemented through a lookup table, processing the four remote command states (such as on / off, mode switching) into 2-bit codes. The specific mapping rules are predefined by the system.

[0084] Let be the state signal vector synchronously read at time t, where It is a fault signal (PB22 pin). It is a low-order signal (PB23 pin). It is a high-order signal (PA27 pin). This is the door lock signal (PA28 pin), each ∈{0,1}. This vector is generated synchronously within a 10ms sampling period.

[0085] Let be the state function, calculated as The value range is {0,1,…,15}. This function compresses the four-dimensional state vector into a 4-bit state word, which directly corresponds to the binary combination of the state signals.

[0086] The shunt resistor voltage value (in volts) acquired by the ADC channel at time t is obtained through shunt resistor measurement of the ULN2803 drive circuit. This voltage value is acquired at a sampling rate of 500Hz and converted to a current value based on a pre-stored resistance-current mapping table, but the voltage value is used directly as input in this formula.

[0087] Let be a function of current, calculated as The range of values ​​is {0, 1, ..., 255}. This is the ADC full-scale voltage (unit: volts), corresponding to the maximum load current. (Unit: Ampere), and , This is the shunt resistor value (unit: ohms). This function quantizes the voltage value into an 8-bit integer, representing the load current timing sequence.

[0088] It should be noted that the washing machine receives remote start commands (12V high-level signal) and pause commands (0V low-level signal).

[0089] Signal conversion: PA08 pin: Receives 12V "Start" signal → COSMO1010 optocoupler conversion → Outputs 3.3V TTL level (logic 1); PA09 pin: Receives 0V "Pause" signal → Outputs 0V TTL level (logic 0); PA10 pin: Receives 12V "Mode Switch" signal → Outputs 3.3V TTL level (logic 1); PA11 pin: Receives 0V "Reserve" signal → Outputs 0V TTL level (logic 0);

[0090] Generate instruction set: Original vector: = (1, 0, 1, 0); Applying a remote function: = Lookup table (1,0,1,0) = 2 (binary "10"); Standard remote instruction set: "10" (represents a start + mode switch compound instruction);

[0091] Status detection signals are acquired synchronously, with 10ms synchronous sampling: PB22 pin (fault signal): 0 (no fault); PB23 pin (low-level signal): 1 (water level is below threshold); PA27 pin (high-level signal): 0 (water level is within limit); PA28 pin (door lock signal): 1 (door is locked).

[0092] Generate state vector: = (0, 1, 0, 1); Apply the state function: = 5 (binary "0101");

[0093] Load current dynamic quantization, ADC sampling: shunt resistor value: = 0.1Ω; Maximum load current: =10A → = 10A × 0.1Ω = 1.0V; Current sampling voltage: = 0.32V (corresponding to an actual current of 3.2A);

[0094] Current processing: Applying current functions: = 82; Binary processing: "01010010";

[0095] Timestamp alignment and fusion, time synchronization: System timestamp: T = 1520ms (millisecond precision);

[0096] Circular buffer alignment: Write buffer index: 0x2000F010; Merged data packets: {T:1520, :2, :5, :82};

[0097] Data dimensionality compression processing, final synthesis: according to formula Calculate: 2×4096 + 5×256 + 82 = 8192 + 1280 + 82 = 9554;

[0098] Binary representation: 0010 0101 01010010; Stored in SRAM at address: 0x2000F000.

[0099] Specifically, the phase space point set reconstruction involves extracting the load current time sequence from the multi-dimensional device state dataset, sampling the current sequence at equal intervals using a preset time delay parameter, constructing a three-dimensional phase space vector set, and generating a reconstructed phase space point set; adjacent trajectory distance calculation involves locating the nearest neighbor vector pair in the reconstructed phase space point set, calculating the ratio of the initial distance to the evolved distance, and generating a distance ratio sequence; Lyapunov exponent fixed-point calculation involves performing logarithmic operations on the distance ratio sequence using the floating-point unit of the main control chip, accumulating the calculation results using a sliding window, and outputting the Lyapunov exponent value; dual-threshold chaos determination involves comparing the Lyapunov exponent value with preset positive and negative thresholds: if it is greater than the positive threshold, output a chaos flag (1); if it is less than the negative threshold, output a stability flag (0); otherwise, maintain the previous state flag, generate a 1-bit chaos characteristic determination flag, and store it in a register.

[0100] It should be noted that the current harmonic analysis involves real-time acquisition of the washing machine's main circuit current signal via a current sensor, decomposing the fundamental frequency (50Hz) and the 3rd, 5th, 7th, and 9th odd harmonic components: fundamental frequency threshold: ≥7.0A (indicating high load status); 3rd harmonic: 1.4–1.6A (activating the "washing status" indicator); 5th harmonic: 0.9–1.1A (activating the "rinsing status" indicator). The harmonic combination characteristics can distinguish between stages such as water injection, washing, and spin-drying.

[0101] Mechanical vibration monitoring: Four displacement sensors (inductive / capacitive) are installed in the washing tub suspension system to detect the three-dimensional vibration offset of the outer drum in real time. Weighing stage: The weight of the clothes is calculated by weighting the displacement (displacement D=2.5mm corresponds to a load of 3kg). Spin-drying stage: If the vibration amplitude exceeds the threshold (>5mm), an imbalance alarm is triggered.

[0102] Optical turbidity detection: Infrared light-emitting diodes and phototransistors on both sides of the drain pipe constitute a transmittance sensor. The water transparency is judged by the rate of change: mud and dirt: transmittance decreases rapidly (30% within 10 seconds); oil and dirt: transmittance decreases slowly (10% within 10 seconds).

[0103] Chaotic dynamics characteristic calculation and phase space reconstruction: Based on the load current time sequence (sampling rate 500Hz), with a preset time delay τ=5 and embedding dimension m=3, a three-dimensional phase space point set is constructed.

[0104] Example: Current sequence [1.8A, 1.9A, 2.0A, 2.1A, 1.7A...] → Vectorized to generate (1.8, 1.9, 2.0), (1.9, 2.0, 2.1), etc.

[0105] Lyapunov exponent calculation: locate the nearest neighbor vector of the current state point in phase space (Euclidean distance 0.42A); track the distance change after 5 evolution steps (distance ratio 2.02); calculate the logarithmic mean (ln(2.02)≈0.70) using floating-point units, and output the Lyapunov exponent value (0.65).

[0106] Two-layer Bayesian state reasoning constructs a probabilistic reasoning network to transform physical signals into state decisions: Upper-layer feature nodes: six types of feature nodes including current harmonics (fundamental wave, 3rd / 5th / 7th / 9th harmonics), turbidity change rate, vibration amplitude, etc., are activated according to thresholds (fundamental wave ≥ 7.0A → node "activated").

[0107] Lower-level state nodes: 4 states: water injection, washing, rinsing, and dehydration. The probability is calculated using the conditional probability formula: Example: If the fundamental wave node is activated + the 3rd harmonic node is activated + the turbidity change rate is high → the probability of the washing state is ≥85%.

[0108] Adaptive parameter closed-loop optimization and dynamic threshold correction: Adjust the Lyapunov exponent threshold based on historical fault counts: Fault count > 10 times: threshold amplified by 120% (increases chaos sensitivity); Fault count < 5 times: threshold reduced by 80% (reduces false alarm rate).

[0109] Proportional Factor Learning: The turbidity sensor updates the proportional factors of dirt properties (δ1= rate of change of mud and δ2= rate of change of oil) through short-term learning, and achieves long-term calibration by combining user manual settings (fabric type, degree of dirt).

[0110] The comparison is shown in Table 4:

[0111] Table 4 Comparison of Recognition Dimensions

[0112] Identification Dimensions physical sensors signal characteristics Identification mechanism Current harmonics Current transformer Fundamental / Odd Harmonic Amplitude Threshold activation Bayesian nodes Mechanical vibration Displacement sensor (4 directions) Weighted displacement / amplitude Over-threshold alarm water turbidity Infrared transmittance sensor Transmittance change rate Adaptive scaling factor Chaotic properties Current sequence phase space reconstruction Lyapunov index Dual threshold determination (chaotic / stable)

[0113] 103. Processing module, used for predictive control instruction generation: When the chaotic characteristic determination flag indicates chaotic characteristics: predict the future working state based on the reconstructed phase space and generate fault prediction state data; if the fault prediction state data indicates an abnormality, trigger the channel switching instruction set through the PB12-PB15 pins to generate predictive protection control instructions.

[0114] Specifically, multi-parameter fusion prediction: In the reconstructed phase space point set, the five nearest neighbors of the current state point are selected. Combining the high-order state signal of pin PA27 and the door lock state signal of pin PA28, a multi-dimensional evolution model is fitted to predict the relay contact impedance change trend after 300 seconds, generating a predicted contact state value; the predicted contact state value is set as follows: :

[0115] ;

[0116] in, Let t be the state variable at the current time. Let be the state variables at historical time t-τ. For time delay; The state variables are for an earlier historical moment t-2τ.

[0117] Dynamic threshold status processing: Obtain dynamic threshold parameters and compare the predicted contact status value with the dynamic threshold: If the predicted value exceeds the impedance safety threshold, output "10" (contact abnormality); if the predicted value is lower than the conduction reliability threshold, output "01" (conduction risk); if the predicted value is within the normal range, output "00" (normal state); generate 2-bit fault prediction status data; predictive power failure protection: When the fault prediction status data is "10" or "01": cut off the DC12V control power supply of the abnormal channel through the ULN2803 driver, activate the enable signal of the backup channel through the PA20 pin, and generate a channel switching instruction set; status feedback optimization: read the current value of the fault counter through the PB22 pin, dynamically adjust the parameter weights of the prediction model according to the fault count, and generate adaptive prediction parameters; protection instruction synthesis: combine the channel switching instruction set with the adaptive prediction parameters, add a 32-bit timestamp and a 16-bit device identifier, and generate predictive protection control instructions;

[0118] It should be noted that when the washing machine enters the spin-drying stage, abnormal vibration is detected (chaos flag = 1). Data preparation: reconstructed phase space point set: 196 three-dimensional vectors constructed based on the load current sequence; current state point: (2.1, 1.9, 2.3) unit: A; 5 nearest neighbor points: (2.0, 2.2, 2.1), (2.2, 2.0, 2.4), (1.9, 2.3, 2.2), (2.3, 1.8, 2.5), (2.0, 2.1, 2.4); state signals: PA27 (high bit) = 0 (water level normal), PA28 (door lock) = 1 (locked).

[0119] Impedance prediction: using a weighted evolution model. ;

[0120] The coefficients (0.82, 0.15, 0.03) represent the weighting factors indicating the degree of contribution of different historical time periods to the predicted value.

[0121] 0.82 represents the current state ( The weight (dominant role) of )

[0122] 0.15 represents the state at the previous delay time ( The weight (secondary role) of )

[0123] 0.03 represents the state at the first two delay times ( The weight (weak effect) of )

[0124] The calculated predicted value for the relay contact state is 83mΩ (normal range 50-80mΩ).

[0125] Dynamic threshold status handling, threshold parameters: impedance safety threshold: 80mΩ (exceeding this threshold indicates an anomaly); conduction reliability threshold: 55mΩ (below this threshold indicates a risk).

[0126] Status determination: Predicted value 83mΩ > 80mΩ → Output "10" (contact abnormal); Generate 2-bit fault prediction status data: "10";

[0127] Predictive power failure protection, protection action: disconnect the DC12V control power supply connected to pin PB13 (abnormal channel) via ULN2803 driver; activate the enable signal of the backup drive channel via pin PA20; generate channel switching instruction set: {main channel power failure, backup channel activation};

[0128] Status feedback optimization and parameter adjustment: Read the PB22 pin fault counter value: 8 (historical number of faults);

[0129] Adjust the prediction model weights based on the fault count: Current weight: 0.82 (current state) → adjusted to 0.85; Historical weight: 0.15 (t-τ) → adjusted to 0.12; Future option weight: 0.03 (t-2τ) → adjusted to 0.03; Generate adaptive prediction parameters: {0.85, 0.12, 0.03};

[0130] Protection command synthesis and assembly: Channel switching command set: main channel power off + backup channel activation; Adaptive prediction parameters: {0.85, 0.12, 0.03}; Add timestamp: 0x5F2A3B1C (August 20, 2024, 14:35:24); Device identifier: 0x0A37 (washing machine device number);

[0131] Final output: Generates a 56-bit predictive protection control command, in the format: [32-bit timestamp][16-bit device ID][4-bit control command][4-bit parameter identifier].

[0132] When the washing machine experiences chaotic current oscillations due to uneven distribution of clothes during the spin-drying stage, the system predicts that the relay contact impedance will rise to 83mΩ (exceeding the safety threshold) after 300 seconds. The system immediately cuts off the main control channel power, switches to the backup channel, and simultaneously fine-tunes the prediction model parameters based on the historical number of faults (8 times), and packages and sends the protection command. The entire process is completed within 100ms, preventing permanent damage caused by contact overheating.

[0133] 204. Control module for IoT communication optimization transmission: The predictive protection control command is processed into a 1-byte simplified status code and transmitted to the IoT management system through at least one of the following communication interfaces: 4G wireless module controlled by PA00-PB09 pins, Ethernet interface controlled by PB00-PB03 pins, and RS485 bus of PA24-PA25 pins.

[0134] Specifically, the instruction bit field compression involves extracting the control field (4 bits) and cooling field (4 bits) from the predictive protection control instruction and merging them into an 8-bit original control word; dynamic time window processing involves obtaining the lower 4 bits of the current system timestamp as a time identifier, performing an XOR operation between the higher 4 bits of the 8-bit original control word and the time identifier to generate a dynamic check code; simplified status code synthesis involves writing the dynamic check code into the lower 4 bits, retaining the lower 4 bits of the 8-bit original control word, and combining them to generate a 1-byte simplified status code; and utilizing multi-protocol adaptive transmission, including: 4G wireless transmission: sending the status code to W via the PA00-PA03 pins. The H-GM5 module simultaneously activates the PB08 pin as a transmit enable signal to generate 4G wireless data packets; Ethernet transmission: the status code is encapsulated into a UDP single-frame message, and the CH9120 chip is driven through the PB00-PB03 pins to generate an Ethernet single-frame message; RS485 transmission: the status code is converted into Modbus RTU format through the PA24-PA25 pins, a device address prefix is ​​added, and an RTU command frame is formed; transmission anomaly handling: when no confirmation is received from the IoT platform after 3 consecutive transmissions, the status code is backed up and stored in the EEPROM through the PB07 pin to generate an offline storage record.

[0135] It should be noted that the instruction bit field is compressed, and the input data is the predictive protection control instruction (56 bits) generated from step 203.

[0136] Field extraction: Control field (4 digits): "1010" (indicates channel switching + parameter update); Cooling field (4 digits): "0011" (indicates medium cooling requirements);

[0137] Merge operation: Merge the control code and cooling code into an 8-bit original control word → "10100011" (0xA3);

[0138] Dynamic processing of the time window to obtain the time identifier: Current system timestamp: 0x5F2A3B1C (August 20, 2024, 14:35:24); Take the lower 4 bits: 0xC (binary "1100");

[0139] XOR operation: The high 4 bits of the original control word are "1010"; XOR the time identifier "1100" with "1010" → "1100" = "0110"; Generate dynamic checksum: "0110";

[0140] Simplified status code synthesis and generation: retain the lower 4 bits of the original control word: "0011"; write the dynamic check code into the lower 4 bits (replacing the original lower 4 bits in actual operation); the final 1-byte simplified status code: "01100011" (0x63).

[0141] Multi-protocol adaptive transmission, 4G wireless transmission: Send status code 0x63 to the WH-GM5 module via PA00-PA03 pins; activate PB08 pin (send enable signal) to generate a high-level pulse;

[0142] Ethernet transmission: The CH9120 chip is driven via pins PB00-PB03;

[0143] Encapsulated as a UDP single-frame packet: Destination IP: 192.168.1.100 (IoT platform); Payload: | 0x63 | Timestamp 0x5F2A3B1C |;

[0144] RS485 transmission: Converted to Modbus RTU format via PA24-PA25 pins; Command frame: | Device address 0x01 | Function code 0x10 | Data 0x63 | CRC16 0xB4A1 |;

[0145] Transmission error handling, trigger condition: three consecutive transmissions without acknowledgment (timeout time 500ms / transmission).

[0146] Backup operation: Trigger a high level via the PB07 pin to start EEPROM storage; Storage format: | Timestamp 0x5F2A3B1C | Status code 0x63 | Retry count 0x03 |; Generate offline storage records to address 0x0000-0x0003;

[0147] When the washing machine predicts a relay contact abnormality (impedance 83mΩ), it generates a protection command and compresses it into a 1-byte status code 0x63. The system simultaneously transmits this status code through three channels: 4G, Ethernet, and RS485. 4G transmission takes 120ms with a data packet size of 12 bytes; Ethernet transmission takes 15ms with a UDP packet size of 28 bytes; and RS485 transmission takes 45ms with an RTU frame size of 6 bytes. If the IoT platform does not respond within 1.5 seconds, status code 0x63 will be backed up to the EEPROM to ensure that data is not lost after a power outage.

[0148] 105. Update module, used for dynamic parameter closed-loop configuration: Receive control parameters issued by the IoT management system, update the Lyapunov exponent threshold in step 202, generate an adaptive parameter set and feed it back to the chaotic feature mathematical operation module.

[0149] Specifically, multi-protocol parameter parsing: Modbus RTU format parameter packets are received via the RS485 interface on pins PA24-PA25, or UDP parameter packets are parsed via the Ethernet interface on pins PB00-PB03, or transparent parameter data is obtained via the 4G module on pins PA00-PA03. The original hexadecimal parameter stream is extracted to generate a parameter data packet; Threshold separation and verification: The high 8 bits of the Lyapunov positive threshold, the low 8 bits of the negative threshold, and the 2-bit check flag are separated from the parameter data packet. The CRC check module is triggered via pin PB06 to verify data integrity and generate a threshold parameter group that has passed verification; Environmental adaptive correction: Read the current value of the fault counter connected to the PB22 pin, and dynamically scale the threshold parameter group according to the fault count level: when the fault count is >10, the threshold is amplified by 120%; when the fault count is <5, the threshold is reduced by 80%, generating an environmental correction threshold parameter; parameter set encapsulation: combine the environmental correction threshold parameter with the device operating mode flag (PA20 pin status), add a version number and timestamp identifier, and generate an adaptive parameter set; closed-loop feedback writing: write the adaptive parameter set to the starting address 0X20002000 of SRAM, send a parameter reception confirmation signal to the IoT management system through the PB10 pin, and generate a parameter update completion flag.

[0150] It should be noted that when the IoT platform issues a Lyapunov index threshold update command, data reception is performed via the RS485 interface on pins PA24-PA25, receiving Modbus RTU format parameter packets; data frame: | Device address 0x01 | Function code 0x06 | Data 0x0F3C | CRC16 0x8A44; | Raw hexadecimal parameter stream: 0x0F3C (representing a positive threshold of 0.60 and a negative threshold of -0.36);

[0151] Threshold separation and verification, parameter separation: high 8 bits: 0x0F → positive threshold = 0.60; low 8 bits: 0x3C → negative threshold = -0.36; verification flag: 0x00 (default);

[0152] CRC verification: The CRC verification module is triggered via the PB06 pin; the CRC16 value of the received data is calculated to be 0x8A44, which matches the frame tail; a threshold parameter group for successful verification is generated: {positive threshold: 0.60, negative threshold: -0.36};

[0153] Environmental adaptive correction, fault count reading: Read the fault counter value of PB22 pin: 12 (historical number of faults); Fault level: >10 (high fault frequency);

[0154] Dynamic scaling: According to the rule: when the fault count > 10, the threshold is amplified by 120%; Correction calculation: positive threshold: 0.60 × 120% = 0.72; negative threshold: -0.36 × 120% = -0.432; generation environment correction threshold parameters: {0.72, -0.432};

[0155] Parameter set encapsulation, device status acquisition: Read PA20 pin status: 1 (washing machine is in energy saving mode);

[0156] Parameter combination: Environmental correction threshold parameters: {0.72, -0.432}; Device operation mode flag: 0x01 (energy saving mode); Version number: 0x02 (second version parameter); Timestamp: 0x5F2A3B1C (current time);

[0157] Generate adaptive parameter set: 32-bit data packet: | 0.72 (16-bit) | -0.432 (16-bit) | Mode + Version (8-bit) | Reserved (8-bit) |;

[0158] Closed-loop feedback write, SRAM storage: Write address: 0x20002000; Data content: 32-bit adaptive parameter set; Storage format: Big-endian mode;

[0159] Confirmation feedback: Sends an acknowledgment signal to the IoT platform via the PB10 pin; Pulse width: 50ms high level; Generates parameter update completion flag: 0xAA55;

[0160] When the IoT platform issues a new threshold parameter (0.60 / -0.36), the system automatically amplifies the threshold by 20% based on the washing machine's current fault count (12 times), resulting in a corrected value (0.72 / -0.432). The new parameter, combined with the device's operating mode (energy-saving mode), is stored in SRAM and a confirmation signal is sent back. This process allows the Lyapunov index threshold to automatically adjust according to the device's aging: the judgment standard is relaxed when the fault frequency is high (to avoid oversensitivity), and the standard is tightened when the fault frequency is low (to improve detection accuracy).

[0161] In this embodiment of the invention, multiple physical signal characteristics such as current harmonic analysis, mechanical vibration monitoring, and optical turbidity detection are combined. Through two-layer Bayesian state reasoning, physical signals are transformed into state judgments, improving the accuracy of fault identification. The Lyapunov exponent judgment threshold is dynamically adjusted based on historical fault counts. The fouling property proportional factor is updated through proportional factor learning, realizing adaptive closed-loop optimization of parameters. This enhances the system's adaptability to different operating conditions and equipment aging levels, enabling early detection of potential chaotic characteristics in equipment, prediction of fault occurrence, and time to take preventive measures. This effectively reduces equipment failure rate, maintenance costs, and production losses.

[0162] By selecting nearest neighbor points from the reconstructed phase space point set and fitting a multidimensional evolution model with other state signals, the future working state is predicted, improving the accuracy of the prediction. Based on the comparison between the predicted value and the dynamic threshold, fault prediction state data is generated to accurately determine the equipment status and provide a reliable basis for the generation of control commands. When the fault prediction state data indicates an abnormality, the channel switching command set is quickly triggered through a specific pin to generate predictive protection control commands, cut off the power supply of the abnormal channel, activate the backup channel, and ensure equipment safety. The prediction model parameter weights are dynamically adjusted according to the fault count to generate adaptive prediction parameters, further improving the performance and reliability of the prediction model, realizing predictive maintenance of the equipment, taking measures before the fault occurs, avoiding the expansion of equipment damage, extending the service life of the equipment, and improving the continuity and stability of the production process.

[0163] Key fields are extracted from predictive protection control commands, merged into original control words, and dynamically generated through time windows to synthesize simplified status codes, reducing data transmission volume and improving transmission efficiency. It supports multiple communication interfaces such as 4G wireless, Ethernet, and RS485 bus, automatically selecting the appropriate transmission method based on actual conditions to ensure timely and reliable data transmission to the IoT management system. When multiple consecutive transmissions fail to receive confirmation, a status code backup is triggered and stored in EEPROM, generating an offline storage record to ensure no data loss and enhance system reliability. This enables efficient and reliable communication between devices and the IoT management system, allowing managers to monitor device status in real time, make timely decisions, and improve the intelligence level of device management.

[0164] It supports receiving control parameters from the IoT management system through multiple communication interfaces, extracting the raw parameter stream to generate parameter data packets, improving system compatibility and flexibility. Threshold parameters are accurately separated from the parameter data packets, and data integrity is verified through CRC checksum to ensure the accuracy of received parameters. Threshold parameter groups are dynamically scaled according to fault count levels, enabling the system to automatically adjust parameters based on the actual operating environment and fault conditions, improving system adaptability and robustness. Environmentally corrected threshold parameters are combined with device operating mode flags to generate an adaptive parameter set, which is written to SRAM and a confirmation signal is sent back to the IoT management system, achieving closed-loop management and updating of parameters. This allows the system to automatically adjust parameters according to changes in the external environment and device operating status, maintaining optimal operating performance, improving the system's intelligence and automation, and reducing manual intervention.

[0165] The present invention also provides an intelligent control device, which includes a memory and a processor. The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor performs the steps of the intelligent control system described in the above embodiments.

[0166] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the intelligent control system.

[0167] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0168] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0169] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent control system, characterized in that, The intelligent control system includes: The acquisition module is used to acquire remote control signals, acquire external status detection signals through pins, and sample load current in real time. The identification module is used to generate a multi-dimensional device status dataset based on remote control signals, external status detection signals, and sampled load current. It performs phase space reconstruction mathematical operations on the multi-dimensional device status dataset, calculates the Lyapunov exponent of the load current sequence using the floating-point unit built into the main control chip, outputs a chaotic characteristic judgment flag, and sets the multi-dimensional device status dataset as... ,but: ； in, The TTL level vector of the remote control signal after optical coupler conversion at time t; It is a remote function; Let be the state signal vector synchronously read at time t, where It is a fault signal. It is a low-order signal. It is a high-level signal. It's a door lock signal; Let be the state function, calculated as The range of values ​​is {0,1,…,15}; The shunt resistor voltage value acquired by the ADC channel at time t; Let be a function of current, calculated as ,in This is the ADC full-scale voltage, corresponding to the maximum load current. ,and , This is the shunt resistor value; The load current time sequence is extracted from the multi-dimensional device status dataset. The current sequence is sampled at equal intervals with a preset time delay parameter to construct a three-dimensional phase space vector set and generate a reconstructed phase space point set. In the reconstructed phase space point set, the nearest neighbor set of the current state point is selected. Combined with the high-level state signal and the door lock state signal, a multi-dimensional evolution model is fitted to predict the contact impedance change trend and generate the contact state prediction value. The contact state prediction value is set to : ； in, Let t be the state variable at the current time. For a historic moment State variables, For time delay; For an earlier historical moment The state variable; 0.82 represents the current state. The weight; 0.15 represents the state at the previous delay time. The weight; 0.03 represents the state at the first two delay times. The weights; The processing module is used to generate fault prediction state data based on the reconstructed phase space to predict the future working state when the chaotic characteristic determination flag indicates chaotic characteristics. The control module is used to trigger a channel switching instruction set, generate a predictive protection control instruction, and transmit the predictive protection control instruction to the controlled device if the fault prediction status data indicates an abnormality.

2. The intelligent control system according to claim 1, characterized in that, include: Locate the nearest neighbor vector pairs in the reconstructed phase space point set, calculate the ratio of the initial distance to the evolved distance, and generate a distance ratio sequence; The distance ratio sequence is logarithmically calculated by the floating-point unit of the main control chip, and the result is accumulated by a sliding window to output the Lyapunov exponent value. Compare the Lyapunov exponent value with a preset positive and negative threshold: if it is greater than the positive threshold, output a chaos flag; if it is less than the negative threshold, output a stability flag; otherwise, maintain the previous state flag and generate a chaos characteristic determination flag.

3. The intelligent control system according to claim 2, characterized in that, include: Obtain dynamic threshold parameters, compare the predicted contact state value with the dynamic threshold, and generate fault prediction state data; A channel switching instruction set is generated based on the fault prediction status data; Read the current value of the fault counter, dynamically adjust the parameter weights of the prediction model based on the fault count, and generate adaptive prediction parameters; The channel switching instruction set is combined with adaptive prediction parameters, and a device identifier is added to generate predictive protection control instructions.

4. The intelligent control system according to claim 1, characterized in that, include: Extract the control field and cooling field from the predictive protection control command and merge them into an 8-bit original control word; Obtain the lower 4 bits of the current system timestamp as a time identifier, and perform an XOR operation between the higher 4 bits of the 8-bit original control word and the time identifier to generate a dynamic check code. Write the dynamic check code into the lower 4 bits, retain the lower 4 bits of the 8-bit original control word, and combine them to generate a 1-byte simplified status code. Utilizing multi-protocol adaptive transmission, when no confirmation is received from the IoT platform after three consecutive transmissions, a status code backup is triggered and stored in EEPROM, generating an offline storage record.

5. The intelligent control system according to claim 1, characterized in that, It also includes an update module, used to receive control parameters from the IoT management system, update the Lyapunov exponent threshold, and generate an adaptive parameter set. Receive Modbus RTU format parameter packets through RS485 interface, parse UDP parameter packets through Ethernet interface, or obtain transparent parameter data through 4G module, extract the original hexadecimal parameter stream, and generate parameter data packets. The verification flag is separated from the parameter data packet, triggering the CRC verification module to verify data integrity and generate a threshold parameter group that passes the verification. Read the current value of the fault counter, dynamically scale the threshold parameter group according to the fault count level, and generate environmental correction threshold parameters; The environmental correction threshold parameter is combined with the device operating mode flag, and a version number and timestamp are added to generate an adaptive parameter set. The adaptive parameter set is written to the starting address 0x20002000 of the SRAM, a parameter reception confirmation signal is sent to the IoT management system, and a parameter update completion flag is generated.

6. The intelligent control system according to claim 5, characterized in that, The verification flag includes: a high 8-bit Lyapunov positive threshold, a low 8-bit negative threshold, and 2 bits; The dynamic scaling of the threshold parameter group according to the fault count level includes: When the fault count is >10, the threshold is amplified by 120%; When the fault count is less than 5, the threshold is reduced by 80%.

Images