Ims-based intelligent early warning communication method for power plant

Abstract

The application relates to the technical field of industrial communication, and discloses a power plant intelligent early warning communication method based on IMS, which comprises the following steps: acquiring discrete structured parameters and continuous flow media sequences, distributing system time stamps and three-dimensional physical coordinate output basic fusion data sequences, and establishing memory mapping logic with equipment asset coding; calculating the numerical deviation degree of characteristic values, combining historical fault characteristic matrix to perform mode matching operation to generate a fault level identifier; activating a hardware timer to generate a control right transfer instruction when a response signal is not captured; constructing a session initiation protocol data packet to be sent to an IP multimedia subsystem network to establish an audio and video transmission channel; evaluating network connectivity, triggering a link switching interruption when communication blockage is detected, and extracting a mapping short code to be sent by external satellite communication hardware. The application realizes data space-time alignment, establishes a transmission channel under normal and blocked working conditions, and maintains communication continuity.

Description

Technical Field

[0001] This invention relates to the field of industrial communication technology, specifically to an intelligent early warning communication method for power plants based on IMS. Background Technology

[0002] The operation monitoring system of power plants relies on a large number of sensors and monitoring terminals deployed on site. With the application of IP multimedia subsystems in industrial communication, voice, video and control commands within power plants are gradually merging into a unified network. However, in complex industrial environments, existing early warning communication systems have technical limitations in terms of data processing depth and link assurance under extreme conditions.

[0003] Existing monitoring systems lack a strict alignment mechanism based on underlying system time and three-dimensional physical coordinates when processing discrete structured sensor parameters and continuous streaming video data. This results in low efficiency in the fusion processing of multi-source heterogeneous data, causing the assessment of abnormal states to deviate from objective quantitative standards. Furthermore, after triggering an early warning state, the existing system's handling process relies excessively on human-computer interaction confirmation by on-site personnel. In situations where sudden emergencies prevent personnel from operating terminals in a timely manner, the lack of automatic transfer logic combining underlying hardware timers causes the early warning process to stall, delaying subsequent resource scheduling.

[0004] Regarding the underlying communication network architecture, when a power plant experiences an extreme failure that causes a physical disruption of the backbone wired network, existing equipment lacks an interruption mechanism to detect network connectivity and directly trigger hardware-level communication link switching. In addition, the audio and video alarm data packets in a conventional IP multimedia subsystem network environment are massive in volume. When the backup satellite communication hardware is activated, due to the lack of established lightweight data mapping and extraction rules for narrowband channels, the original large-volume data payload exceeds the bandwidth limit of satellite transmission, resulting in the failure to successfully send core fault parameters. Consequently, the remote service takeover platform cannot obtain the on-site fault status in a timely manner. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent early warning communication method for power plants based on IMS, which solves the problems of low efficiency in heterogeneous data fusion processing in existing power plants and the lack of link switching and lightweight data transmission mechanisms when the backbone communication network is blocked, resulting in alarm information not being able to be reported to the remote platform.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A smart early warning communication method for power plants based on IMS, comprising the following steps: Obtain discrete structured parameters and continuous streaming media sequences, assign system timestamps and three-dimensional physical coordinates to the discrete structured parameters and the continuous streaming media sequences, and combine and output basic fusion data sequences to establish memory mapping logic between the basic fusion data sequences and device asset codes; Feature values ​​are extracted from the basic fusion data sequence, the deviation of the feature values ​​from the preset operating benchmark indicators is calculated, and the deviation is used to perform pattern matching operation with the preset historical fault feature matrix to generate a fault level identifier. Based on the fault level identifier, output local environmental sound and light control instructions and activate hardware timers. If the hardware timer reaches the set over-limit time threshold and no response signal from the human-machine interaction terminal is captured, generate a control transfer instruction. The system receives the control transfer instruction to extract the target communication address, constructs a session initiation protocol data packet with a custom extended field containing the fault level identifier inserted in the packet header, and sends the session initiation protocol data packet to the IP multimedia subsystem network to establish an audio and video transmission channel. The network layer probe message is sent to the IP Multimedia Subsystem network to assess network connectivity. When communication blocking characteristics are detected, the underlying communication link switching interruption is triggered. The mapping short code corresponding to the basic fusion data sequence and the fault level identifier is extracted and sent using external satellite communication hardware.

[0007] By allocating the discrete structured parameters with the timestamps and three-dimensional physical coordinates of the continuous streaming media sequence, a spatiotemporal alignment benchmark for multi-source data is established; pattern matching calculations are performed using numerical deviation and historical fault feature matrices to output objective and quantitative fault classification results; the response status of the human-machine interface terminal is monitored by a hardware timer, triggering the transfer of control under unattended conditions; when network connectivity assessment determines communication blockage, the mapped short code is extracted and transmitted using external satellite communication hardware to establish a backup alarm data transmission channel under limited bandwidth.

[0008] Preferably, the steps of obtaining discrete structured parameters and continuous streaming media sequences, assigning system timestamps and three-dimensional physical coordinates to the discrete structured parameters and the continuous streaming media sequences, combining and outputting a basic fused data sequence, and establishing memory mapping logic between the basic fused data sequence and device asset codes include: The discrete structured parameters are divided according to a set sampling period to form a discrete data payload, and the continuous image data of the continuous streaming media sequence is divided according to a set time window to form a streaming media data payload. Configure a unified data header containing the underlying hardware media access control address for the discrete data payload and the streaming media data payload, and combine them to generate the basic fused data sequence; The device asset code is obtained by performing an address matching operation using the three-dimensional physical coordinates and the underlying hardware media access control address. Construct a memory address hash mapping table with the device asset code as the retrieval key and the starting logical address of the basic fusion data sequence storage area as the mapping value.

[0009] In the above steps, a basic fused data sequence is generated by configuring a unified data header for discrete data payloads and streaming media data payloads, thereby eliminating the data format differences caused by heterogeneous communication protocols; a memory address hash mapping table with device asset codes as retrieval keys is constructed to shorten the memory addressing cycle of the starting logical address of the basic fused data sequence.

[0010] Preferably, the step of extracting feature values ​​from the basic fused data sequence and calculating the deviation of the feature values ​​from the preset operating benchmark index includes: Extract the feature values ​​from the basic fused data sequence to construct a one-dimensional feature vector; Read the preset operating benchmark index, which includes theoretical expected values ​​and tolerance range parameters; Calculate the difference between the feature value and the corresponding theoretical expected value, and determine the ratio of the difference to the corresponding tolerance range parameter as the value deviation. Combine these to generate a deviation feature vector.

[0011] In the above steps, a preset operating benchmark index containing tolerance range parameters is introduced to calculate the ratio, the absolute difference is converted into a relative numerical deviation, the physical dimensions of different sensor values ​​are unified, and a standardized deviation feature vector is generated as the input parameter for pattern matching operation.

[0012] Preferably, the step of generating a fault level identifier by performing pattern matching operation using the numerical deviation and a preset historical fault feature matrix includes: Calculate the weighted Euclidean distance between the deviation feature vector and the fault center vector in the historical fault feature matrix; The smallest calculated result from the weighted Euclidean distances is determined as the target matching distance, and the fault type corresponding to the target matching distance is recorded as the target fault type. When the target matching distance is less than or equal to the matching acceptance threshold, the fault level code pre-associated with the target fault type is read to generate the fault level identifier.

[0013] In the above steps, the weighted Euclidean distance between the deviation feature vector and the fault center vector in the historical fault feature matrix is ​​calculated to quantify the spatial similarity between the operating state and the known fault model; a matching acceptance threshold is set to block false alarms caused by slight numerical fluctuations, and then the corresponding fault level identifier is output.

[0014] Preferably, the step of generating a control transfer instruction when the hardware timer reaches a set over-time threshold and no response signal from the human-machine interface terminal is captured includes: The over-limit time threshold is obtained based on the fault level identifier and injected into the hardware timer. The stepping operation of the internal register is executed using the system clock source. Start a non-blocking background listening thread to continuously read the interface status register of the human-computer interaction terminal to monitor the action feedback status; When the remaining time of the hardware timer reaches zero and no valid low-level pulse corresponding to the human-machine interaction terminal response signal is captured, the device asset code, the numerical deviation, and the fault level identifier are extracted to generate the control transfer instruction.

[0015] In the above steps, an over-limit time threshold is dynamically injected into the hardware timer based on the fault level identifier. The non-blocking background listening thread directly reads the level status of the interface status register, avoiding thread blocking caused by software polling, and ensuring that the control transfer instruction is generated in real time at the moment of response timeout.

[0016] Preferably, the steps of receiving the control transfer instruction to extract the target communication address and constructing a session initiation protocol data packet with a custom extended field loading the fault level identifier inserted in the packet header include: The target communication address is extracted by querying the responsibility object mapping table based on the fault level identifier; The numerical deviation and the equipment asset are encoded into a waveform audio file; Write session description protocol data containing the media transmission port into the message body of the session initiation protocol data packet, and insert the custom extended field containing the fault level identifier into the header of the session initiation protocol data packet.

[0017] In the above steps, a custom extended field carrying the fault level identifier is written into the header of the session initiation protocol data packet to expose the fault priority status to the network routing node; at the same time, the numerical deviation and device asset encoding are converted into waveform audio files to provide a carrier for directly sending alarm data in the media transmission port.

[0018] Preferably, the step of sending the Session Initiation Protocol (SIP) data packet to the IP Multimedia Subsystem network to establish an audio / video transmission channel includes: Based on the fault level identifier, the IMS signaling priority parameter is determined, and the session initiation protocol data packet is sent using the IMS signaling priority parameter to perform signaling queuing and routing forwarding; After the session confirmation is completed, the waveform audio file and the corresponding video data segment in the continuous streaming media sequence are respectively encapsulated into an audio real-time transmission protocol data stream and a video real-time transmission protocol data stream. The audio real-time transmission protocol data stream and the video real-time transmission protocol data stream are sent respectively according to the agreed media transmission port to establish the audio and video transmission channel.

[0019] In the above steps, the queuing and routing order of the session initial protocol data packets is controlled by the IMS signaling priority parameter, and the fault alarm signaling network is given priority passage rights; the audio real-time transmission protocol data stream and the video real-time transmission protocol data stream are sent separately through the media transmission port to complete the concurrent transmission of live data.

[0020] Preferably, the step of sending a network layer probe message to the IP Multimedia Subsystem network to assess network connectivity and triggering a lower-level communication link handover interruption when communication blocking characteristics are detected includes: Send the network layer probe message to the IMS network connectivity probe address and capture the response data frame to obtain round-trip delay parameters and packet loss rate parameters; The round-trip delay parameter and the packet loss rate parameter are normalized according to the calibrated upper limit value, respectively. The network connectivity evaluation value is calculated using the normalized parameters. When the network connectivity evaluation value is greater than the communication blocking characteristic threshold, a blocking signal is generated and injected into the external interrupt request pin to trigger the underlying communication link switching interruption.

[0021] In the above steps, the network connectivity evaluation value is calculated based on the round-trip delay parameter and packet loss rate parameter obtained from the captured response data frame. When the network connectivity evaluation value exceeds the blocking characteristic threshold, a blocking signal is injected into the external interrupt request pin, and the underlying communication link is switched using the hardware-level interrupt mechanism.

[0022] Preferably, the step of extracting the mapping short code corresponding to the basic fusion data sequence and the fault level identifier includes: During the execution of the exception handling service routine, the exception parameter type is read from the basic fusion data sequence, and the deviation interval code corresponding to the numerical deviation is determined according to the preset deviation interval boundary. The equipment asset code, the abnormal parameter type, the deviation range code, and the fault level identifier are combined to form an alarm feature index; The corresponding mapping short code is extracted from the preset alarm short code hash mapping table based on the alarm feature index.

[0023] In the above steps, abnormal parameter types are read from the basic fusion data sequence and the deviation range code is determined. This code is then combined with the equipment asset code and fault level identifier to form an alarm feature index. The mapping short code is then extracted, and the data volume is compressed while retaining the alarm elements to adapt to the narrowband channel transmission limitations.

[0024] Preferably, the step of transmitting the mapped short code using external satellite communication hardware includes: Write the mapped short code into the transmit buffer register of the external satellite communication hardware; After detecting that the mapped short code to be transmitted exists in the transmit buffer register, the radio frequency transceiver link is activated and link synchronization and channel access are completed; The mapped short code is processed by physical carrier modulation using phase shift keying modulation, and the wireless signal is sent to a low Earth orbit communication satellite node for forwarding to a remote service takeover platform.

[0025] In the above steps, the mapped short code is written into the transmit buffer register of the external satellite communication hardware, the radio frequency transceiver link is activated and physical carrier modulation processing is performed, and a backup physical link is built using the low Earth orbit communication satellite node to complete the remote service takeover under the condition of ground wired communication network interruption.

[0026] This invention provides an intelligent early warning communication method for power plants based on IMS. It has the following beneficial effects: 1. This invention assigns system timestamps and three-dimensional physical coordinates to discrete structured parameters and continuous streaming media sequences, combines them to generate basic fusion data sequences, and establishes memory mapping logic between the basic fusion data sequences and device asset codes. This enables strict alignment of underlying data from different acquisition sources in the time and space dimensions, eliminates format barriers in heterogeneous data processing, and shortens the memory addressing cycle for locating target devices during subsequent anomaly analysis.

[0027] 2. This invention calculates the numerical deviation by extracting the feature values ​​of the basic fusion data sequence, performs pattern matching operations in combination with the preset historical fault feature matrix to generate a fault level identifier, and uses a hardware timer to monitor the response status of the human-machine interaction terminal. It performs fault classification by quantifying the deviation index, eliminating subjective experience interference. At the same time, in conjunction with the underlying hardware timing mechanism, it automatically generates a control transfer instruction when no personnel response level signal is detected, ensuring that the early warning flow mechanism is uninterrupted.

[0028] 3. This invention triggers a low-level link switching interruption when a probe message is sent to the IP Multimedia Subsystem Network and a communication blocking feature is detected. Then, a mapping short code containing equipment assets and fault elements is extracted and sent using external satellite communication hardware. By adopting the design of triggering the low-level interruption and extracting the mapping short code, the bandwidth occupation of alarm data payload can be reduced under the condition of backbone communication network paralysis, thereby meeting the transmission limitations of satellite narrowband channels and establishing an independent data reporting channel. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the system timing interaction of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram of the multi-source data acquisition and basic fusion data processing flow of the present invention; Figure 4 This is a schematic diagram of the intelligent diagnostic classification and pattern matching operation process of the present invention; Figure 5 This is a schematic diagram of the weighted Euclidean distance curve for historical fault mode matching in this invention. Figure 6 This is a schematic diagram of the local alarm response and remote communication scheduling process of the present invention; Figure 7 This is a schematic diagram of the network status detection and satellite communication link switching process of the present invention; Figure 8 This is a schematic diagram of the network connectivity assessment and communication blocking trigger curve of the present invention. Detailed Implementation

[0030] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] See attached document Figure 1 The present invention provides an intelligent early warning communication system for power plants based on IMS, the system comprising: a data acquisition module, a storage module, a diagnostic module, a control module, and a communication module.

[0032] The acquisition module obtains discrete structured parameters output by physical sensors and continuous streaming media sequences output by video surveillance equipment. The acquisition module assigns system timestamps and three-dimensional physical coordinates to the discrete structured parameters and continuous streaming media sequences, thereby generating a basic fusion data sequence. The acquisition module then outputs the basic fusion data sequence to the storage module.

[0033] The storage module is connected to the acquisition module to receive the basic fused data sequence. The storage module establishes a memory mapping logic between the basic fused data sequence and the device asset code according to the set physical device classification dimension, and the storage module provides a data retrieval interface to the outside.

[0034] The diagnostic module connects to the storage module through a data retrieval interface. The diagnostic module extracts the basic fusion data sequence and calculates the numerical deviation of the basic fusion data sequence relative to the operating benchmark index. The diagnostic module uses the numerical deviation to perform pattern matching operation with a preset historical fault feature matrix, and then outputs the corresponding fault level identifier.

[0035] The control module is connected to the diagnostic module to receive fault level identifiers. Based on the fault level identifiers, the control module outputs local environmental sound and light control commands and activates a hardware timer. When the hardware timer reaches the set over-limit time threshold and no response signal is captured from the human-machine interface terminal, the control module generates a control transfer command.

[0036] The communication module is connected to the control module to receive control transfer instructions. The communication module extracts the target communication address based on the fault level identifier and performs a communication signaling format conversion operation to construct a session initiation protocol data packet containing custom extended fields. The communication module sends the session initiation protocol data packet to the IP Multimedia Subsystem network to establish an audio and video transmission channel. When the communication module detects communication interruption characteristics in the IP Multimedia Subsystem network, it triggers a link switching interruption. Then, the communication module generates a mapping short code and drives the external satellite communication hardware to perform information delivery operations in a non-public network environment.

[0037] See attached document Figure 2 This invention provides an intelligent early warning communication method for power plants based on IMS, the method comprising the following steps: S1: Obtain the discrete structured parameters output by the physical sensor and the continuous streaming media sequence output by the video surveillance equipment. Assign system timestamps and three-dimensional physical coordinates to the discrete structured parameters and the continuous streaming media sequence. Merge and package the discrete structured parameters and the continuous streaming media sequence to output the basic fusion data sequence. Establish the memory mapping logic between the basic fusion data sequence and the equipment asset code according to the set physical device classification dimension.

[0038] S2 receives the basic fusion data sequence and extracts the feature values ​​of the basic fusion data sequence, calculates the numerical deviation of the feature values ​​relative to the preset operating benchmark index, retrieves the preset historical fault feature matrix, and performs a pattern matching operation using the numerical deviation and the historical fault feature matrix, and generates the corresponding fault level identifier based on the pattern matching operation result.

[0039] S3 receives the fault level identifier and outputs the local environmental sound and light control command, synchronously activates the hardware timer, and monitors the action feedback status of the human-machine interaction terminal during the running cycle of the hardware timer. If the hardware timer reaches the set over-limit time threshold and no response signal from the human-machine interaction terminal is captured, a control transfer command is generated.

[0040] S4 receives the control transfer instruction, queries the pre-configured responsibility object mapping table based on the fault level identifier, thereby extracting the target communication address, processes the numerical deviation and equipment asset code into waveform audio files through an audio conversion program, constructs a session initiation protocol data packet, inserts a custom extension field into the header of the session initiation protocol data packet, loads the fault level identifier into the custom extension field, and sends the session initiation protocol data packet loaded with the custom extension field to the IP multimedia subsystem network, thereby establishing an audio and video transmission channel pointing to the target communication address.

[0041] S5 sends network layer probe messages to the IP Multimedia Subsystem network and evaluates network connectivity based on the return status of the network layer probe messages. When communication blocking characteristics are detected, a blocking signal is generated, which triggers the underlying communication link switching interruption. The preset alarm short code hash mapping table is read, and the mapping short code corresponding to the basic fusion data sequence and the fault level identifier is extracted. The mapping short code is written into the transmit buffer register of the external satellite communication hardware, and the communication delivery operation in the non-public network environment is performed using the external satellite communication hardware. This completes the closed loop of the entire data processing chain from multi-source data acquisition, intelligent diagnosis and classification to multi-mode communication scheduling.

[0042] See attached document Figure 3 In specific implementations, step S1 provided by the present invention may include the following steps: S11, the acquisition module starts its internal multi-threaded protocol scheduling process, configures the corresponding communication socket resources, and establishes a data transmission link between the acquisition module and the underlying physical sensors and video monitoring equipment. For the unit vibration frequency and dam seepage pressure parameters output by the physical sensors, the acquisition module calls the corresponding industrial communication protocol stack to perform data parsing operations according to the communication protocol used by the underlying physical sensors. When the Modbus-TCP protocol is used, the acquisition module reads the corresponding register value according to the register address; when the OPC-UA protocol is used, the acquisition module accesses the corresponding data node according to the node identifier and reads the numerical attributes of the data node.

[0043] The acquisition module converts the values ​​obtained from different communication protocols into a preset binary data format. The preset binary data format is configured as an unsigned integer data format or a floating-point data format, and the bit width of the preset binary data format is configured as 16 bits or 32 bits, thereby forming discrete structured parameters.

[0044] For the monitoring scene output by the video surveillance equipment, the acquisition module uses the real-time streaming protocol to send a request signal to the hardware port to establish a long network connection and extract continuous image data frames. Then, the acquisition module writes the parsed hexadecimal discrete values ​​and continuous image data frames in parallel in the system memory buffer area, thereby constructing a discrete structured parameter array and a continuous streaming media sequence queue, respectively.

[0045] S12, the acquisition module calls the underlying network time protocol service interface to obtain the absolute system time of the current operating system. The absolute system time is used as the system timestamp of the discrete structured parameter and the continuous streaming media sequence. The obtained absolute system time is represented as a UNIX timestamp data format accurate to milliseconds. At the same time, the acquisition module reads the preset static physical topology configuration file, extracts and parses the corresponding text fields inside the static physical topology configuration file, and extracts the three-dimensional physical coordinates corresponding to each underlying hardware device. The three-dimensional physical coordinates are represented as longitude, latitude and elevation numerical parameters based on the national geodetic coordinate system. For the specific clock offset compensation and round-trip delay calculation of the network time protocol, the time offset parameter can be calculated based on the existing clock synchronization technology. The acquisition module uses the absolute system time as a unified time series reference point and traverses the discrete structured parameter array and the continuous streaming media sequence queue.

[0046] The acquisition module writes the absolute system time and three-dimensional physical coordinates into the header of each discrete data packet of the discrete structured parameter array, and writes the absolute system time and three-dimensional physical coordinates into the header of each key video frame in the continuous streaming media sequence queue. For non-key video frames between adjacent key video frames, the acquisition module records the frame number and the time offset relative to the corresponding key video frame, and determines the system time corresponding to each non-key video frame based on the absolute system time and time offset of the key video frame. The three-dimensional physical coordinates of the non-key video frames are determined according to the three-dimensional physical coordinates corresponding to the video surveillance equipment, thereby completing the spatiotemporal alignment of the discrete structured parameters and the continuous streaming media sequence.

[0047] S13, the acquisition module performs binary packing and encapsulation according to the set bit width alignment and byte padding rules, and the set bit width alignment is performed according to the 32-bit or 64-bit standard. For discrete structured parameter arrays, the acquisition module forms discrete data payloads according to a single sampling period; for continuous streaming media sequence queues, the acquisition module divides continuous image data according to a single video frame or a set time window. The set time window ranges from 500ms to 2000ms, thereby forming multiple streaming media data payloads.

[0048] The acquisition module configures a fixed-length unified data header for both discrete data payloads and streaming media data payloads. The fixed length is configured to be 64 bytes or 128 bytes. The unified data header includes at least the protocol version identifier, data type identifier, payload length identifier, absolute system time, underlying hardware media access control address, and device association identifier. A variable-length payload area corresponding to the actual data length is set after the unified data header.

[0049] The acquisition module generates a basic fused data sequence in the form of a combination of a unified data header and a variable-length payload area. The acquisition module triggers the release and acquisition logic of the memory access control lock through the internal shared memory cross-process communication mechanism, outputs the basic fused data sequence to the write buffer of the storage module and performs access lock restriction operation.

[0050] S14, the storage module receives the basic fused data sequence written to the buffer, extracts the three-dimensional physical coordinates and hardware media access control address from the data packet, and performs address matching operation in the preset physical device classification dimension mapping table using the three-dimensional physical coordinates and hardware media access control address. The number of physical device classification dimensions is configured to be 3 to 5 layers. Then, the storage module matches and obtains the globally unique device asset code associated with the underlying layer. The storage module dynamically allocates a logically contiguous storage area in the system memory heap area and writes the complete basic fused data sequence into the allocated storage area. The storage module constructs a memory address hash mapping table with the device asset code as the retrieval key and the starting logical address of the storage area as the mapping value, thereby establishing the memory mapping logic between the basic fused data sequence and the device asset code, and providing data address query capability with constant time complexity on average.

[0051] See attached document Figure 4 and attached Figure 5 In specific implementations, step S2 provided by the present invention may include the following steps: S21, combined with the heartbeat frequency of the system clock cycle, the diagnostic module connects to and accesses the memory logical address of the storage module through the data retrieval interface. The diagnostic module reads the basic fusion data sequence in the storage area according to the set diagnostic cycle. The set diagnostic cycle ranges from 10ms to 100ms, and the diagnostic cycle is set according to the sampling cycle corresponding to the feature to be diagnosed.

[0052] When different underlying sensing hardware has different sampling frequencies, the diagnostic module selects the data sample that is closest to the current diagnostic time and within the preset allowable time deviation range based on the absolute system time. The preset allowable time deviation range is from ±5ms to ±20ms. Then, the diagnostic module performs time alignment on different data samples to form a one-dimensional feature vector corresponding to the same diagnostic time.

[0053] The diagnostic module strips the unified data header of the basic fused data sequence according to the preset binary message parsing rules, and then extracts the feature values ​​stored inside the discrete structured parameter array. The feature values ​​specifically include, but are not limited to, the operating temperature parameters of physical devices and the bearing vibration amplitude parameters. The diagnostic module performs a matrix splicing operation on the extracted feature values ​​and the absolute system time to construct a one-dimensional feature vector. The diagnostic module then writes the one-dimensional feature vector into the internal high-speed buffer to await the call of the processing unit.

[0054] S22, after loading the one-dimensional feature vector in the high-speed buffer, the diagnostic module reads the preset operating benchmark index in the local memory. The operating benchmark index includes the theoretical expected value and tolerance range parameter under normal operating conditions. The theoretical expected value is represented by the factory calibration parameter of the equipment or the average value of data within a historical fault-free operating cycle. The tolerance range parameter is represented by the maximum allowable deviation value, and the maximum allowable deviation value ranges from 5% to 15% of the theoretical expected value. The diagnostic module uses the internal processing unit to call the numerical deviation calculation formula to calculate the one-dimensional feature vector and the operating benchmark index, thereby obtaining the numerical deviation. The numerical deviation reflects the objective severity of the deviation of the physical quantity of the current real-time operating condition from the standard state. The specific calculation formula is as follows: ; In the formula, For the first The numerical deviation of each feature parameter; This is the ordinal variable for the feature parameter, and its value is a positive integer. To extract the first feature from the obtained one-dimensional feature vector Each feature value; The first of the operating benchmark indicators The theoretical expected values ​​corresponding to each feature parameter; The first of the operating benchmark indicators The tolerance range parameter corresponding to each feature parameter.

[0055] The diagnostic module combines the numerical deviations of all feature parameters to generate a deviation feature vector.

[0056] S23, for the generated deviation feature vector, the diagnostic module retrieves the pre-set historical fault feature matrix in the non-volatile storage medium. The historical fault feature matrix is ​​divided into multiple fault sample sets according to the fault type. Each fault sample set consists of historical deviation sample data of the corresponding fault type. The diagnostic module performs feature-wise statistical operations to calculate the arithmetic mean or the median of the historical deviation sample data in each fault sample set, thereby obtaining the fault center vector corresponding to each fault type, and pre-associating the corresponding fault level code for each fault type.

[0057] The diagnostic module performs pattern matching operations using the current deviation feature vector and the fault center vectors corresponding to different fault types. The diagnostic module calculates the weighted Euclidean distance between the deviation feature vector and the fault center vectors of various types in the historical fault feature matrix through the processing unit. The specific formula for calculating the weighted Euclidean distance is as follows: ; In the formula, The deviation feature vector and the historical fault feature matrix are the first... Weighted Euclidean distance between failure modes; This is a variable representing the category number of historical failure modes, with a value of a positive integer; This represents the total number of characteristic parameters; This is the ordinal variable for the feature parameter, and its value is a positive integer. For the first The decision weight coefficients for each feature parameter range from 0.1 to 0.9. For the first The numerical deviation of each feature parameter; The first in the historical fault feature matrix The first type of failure mode corresponding to the The standard deviation of each feature parameter.

[0058] The diagnostic module determines the smallest calculated result from all weighted Euclidean distances as the target matching distance and records the fault type corresponding to the target matching distance as the target fault type.

[0059] S24, the diagnostic module reads the matching acceptance threshold corresponding to the target fault type. The value range of the matching acceptance threshold is 0.05 to 0.15. Then, the diagnostic module compares the target matching distance with the matching acceptance threshold. When the target matching distance is less than or equal to the matching acceptance threshold, the diagnostic module determines that the current deviation feature vector matches the target fault type and reads the fault level code pre-associated with the target fault type.

[0060] When the target matching distance is greater than the matching acceptance threshold, the diagnostic module marks the current operating status as an unrecorded abnormal condition and assigns it a preset highest-level alarm code. The highest-level alarm code is configured as 0xFF. The diagnostic module generates a fault level identifier based on the fault level code. The fault level identifier is represented as a numeric label containing two hexadecimal digits. The diagnostic module writes the equipment asset code, the numerical deviation of each feature parameter, the fault level identifier, and the absolute system time into the first-in-first-out message queue according to the data structure of the same alarm event, waiting for the control module to read it.

[0061] See attached document Figure 6In specific implementations, step S3 provided by the present invention may include the following steps: S31, the control module establishes a communication connection with the first-in-first-out (FIFO) message queue. Following the FIFO order, the control module extracts the device asset code, the numerical deviation of each characteristic parameter, the fault level identifier, and the absolute system time corresponding to the same alarm event from the FIFO message queue. The reading results are then written to the current alarm event buffer. The control module also reads a preset interface mapping configuration file, which stores pin level control parameters corresponding to different fault levels. The control module parses the hexadecimal code of the fault level identifier and matches it to obtain the corresponding pin level control parameters. The control module outputs corresponding local environmental sound and light control commands to the field alarm devices through a general-purpose input / output interface. These commands are represented as square wave pulse signals with a frequency range of 1Hz to 5Hz. The square wave pulse signals, through high-low level conversion, directly drive the field alarm devices to sound the buzzer and flash the indicator lights.

[0062] S32, when the control module outputs the local environment sound and light control command, it calls the hardware timer interface at the operating system level. The control module synchronously activates the hardware timer based on the hardware timer interface. At the same time, the control module reads the preset timeout mapping table and performs key-value matching calculation in the timeout mapping table according to the fault level identifier to obtain the corresponding over-limit time threshold. Different fault events with different danger levels correspond to different over-limit time thresholds. The over-limit time thresholds are pre-configured in a way that decreases as the fault danger level increases. The higher the danger level of the fault event, the smaller the over-limit time threshold. The value range of the over-limit time threshold is 30s to 120s. Then, the control module injects the over-limit time threshold into the overload register of the hardware timer. The hardware timer executes the stepping operation of the internal register by relying on the pulse beat of the system clock source.

[0063] S33, within the operating cycle of the hardware timer, the control module starts a background listening thread independent of the main process. The background listening thread uses a non-blocking polling mechanism to monitor the action feedback status of the human-machine interface terminal. The specific physical button debouncing and level state machine recognition calculation of the human-machine interface terminal can be handled by existing digital circuit debouncing technology. The control module continuously reads the interface status register of the human-machine interface terminal within the polling cycle, which ranges from 10ms to 50ms. Then, the control module determines whether the human-machine interface terminal response signal has been captured based on the level toggling of the interface status register. The human-machine interface terminal response signal is a valid low-level pulse generated by the operator triggering the confirmation button. If the human-machine interface terminal response signal is captured before the hardware timer reaches the set over-limit time threshold, the control module sends a reset signal to the hardware timer to clear the internal register and terminate the subsequent control transfer process.

[0064] In branch logic S34, where no response signal from the human-computer interaction terminal is captured, the hardware timer performs clock step calculations based on the system clock pulse frequency. The specific formula for calculating the remaining time is as follows: ; In the formula, This represents the remaining time of the hardware timer. The set time threshold for exceeding the limit; This represents the cumulative number of times the system clock pulses have been triggered. The fixed period of a single system clock pulse ranges from 1ms to 10ms.

[0065] When the remaining time reaches zero, the hardware timer sends a hardware interrupt request to the main processor. The control module captures the hardware interrupt request and determines that the hardware timer has reached the set over-limit time threshold. If the set over-limit time threshold is reached and no response signal from the human-machine interface terminal is captured, the control module reads the device asset code, the numerical deviation of each characteristic parameter, the fault level identifier, and the absolute system time from the current alarm event buffer.

[0066] The control module splices and assembles the read data according to a preset communication frame format. The preset communication frame format is configured as a structured data format, such as Extensible Markup Language (EXPLAIN) or JavaScript Object Notation (JavaScript) format. Then, the control module generates a control transfer instruction containing payload encoded data such as message headers and hexadecimal data. The control transfer instruction is used to trigger the alarm handling process to switch from the local confirmation stage to the remote communication dispatch stage. It does not involve the transfer of power generation equipment operation control authority. The control module pushes the generated control transfer instruction into the transmission buffer queue.

[0067] In specific implementations, step S4 provided by this invention may include the following steps: S41, the communication module establishes a low-level memory mapping channel with the control module's sending buffer queue. The communication module receives the control transfer instruction based on the hardware interrupt trigger flag. The communication module uses a message parsing algorithm to extract the fault level identifier encapsulated within the control transfer instruction. The communication module reads the pre-configured responsibility object mapping table built into the non-volatile memory. The pre-configured responsibility object mapping table contains a hash mapping dictionary of different fault levels and corresponding maintenance personnel contact information. Then, the communication module performs a key-value lookup operation in the hash mapping dictionary based on the fault level identifier to extract the target communication address. The target communication address is represented as a session routing endpoint number based on the Uniform Resource Identifier format. The target communication address is stored in a local register as the exact logical destination for subsequent signaling addressing.

[0068] S42, the communication module synchronously parses the numerical deviation and equipment asset code contained in the control transfer instruction. The communication module extracts the numerical deviation and equipment asset code and concatenates them into a structured text string according to a preset syntax structure. Then, the communication module calls the audio conversion program burned into the internal digital signal processor. The specific text-to-speech synthesis logic and acoustic model mapping operation within the audio conversion program can be processed according to existing speech synthesis technology. Furthermore, the communication module uses the audio conversion program to process the structured text string into a waveform audio file. The waveform audio file is in the form of a binary media stream containing pulse code modulation continuous sampling data, and the sampling rate of the continuous sampling data is configured to be 8kHz or 16kHz, thereby forming a continuous data code that converts digital alarm parameters into analog auditory signals.

[0069] S43, the communication module constructs an INVITE request message according to the session initiation protocol state machine specification, and writes the target communication address into the request Uniform Resource Identifier field of the INVITE request message. The communication module writes session description protocol data into the message body of the INVITE request message. The session description protocol data includes at least the audio encoding type, video encoding type, media transmission port, and real-time transmission protocol parameters. The value range of the media transmission port is configured to be from 10000 to 20000.

[0070] The communication module inserts a fault level extension field as a custom extension field into the header of the INVITE request message. The fault level extension field consists of a preset extension field name and a fault level string. The communication module converts the fault level identifier into a string format and writes it into the fault level extension field. After encapsulation, the INVITE request message is written into the network interface buffer queue to wait for transmission.

[0071] S44, the communication module configures the underlying session boundary controller physical network interface. The communication module sends an INVITE request message carrying a fault level extension field through the session boundary controller physical network interface. The session boundary controller pre-stores the mapping relationship between fault level identifiers and IMS signaling priority parameters. The session boundary controller parses the fault level extension field and determines the corresponding IMS signaling priority parameter based on the fault level identifier. The value range of the IMS signaling priority parameter is configured to be from level 1 to level 5. Then, according to the determined IMS signaling priority parameter, the INVITE request message is sent to the IP Multimedia Subsystem Network. The IP Multimedia Subsystem Network completes signaling queuing and routing forwarding according to the IMS signaling priority parameter.

[0072] After the target communication address binding terminal returns a session success response and completes session confirmation, the communication module establishes a media transmission channel pointing to the target communication address. Based on the device asset code and absolute system time contained in the control transfer instruction, the communication module reads the video data segment of the corresponding device near the alarm occurrence time through the data retrieval interface provided by the storage module. The communication module encapsulates the waveform audio file into an audio real-time transmission protocol data stream and the video data segment into a video real-time transmission protocol data stream. Then, according to the media transmission port negotiated by the session description protocol, it sends the audio real-time transmission protocol data stream and the video real-time transmission protocol data stream respectively, thereby establishing an audio and video transmission channel corresponding to the alarm event. At the same time, the communication module calls the short message service gateway interface in the IP multimedia subsystem network, parses the string in the fault level extension field, and synchronously pushes the alarm content to the mobile terminal bound to the target communication address in the form of short message text.

[0073] See attached document Figure 7 and attached Figure 8 In specific implementations, step S5 provided by the present invention may include the following steps: S51, the communication module establishes an interaction channel with the underlying protocol stack of the operating system. The communication module reads the pre-configured IMS network connectivity probe address. The IMS network connectivity probe address corresponds to a session boundary controller network interface or a dedicated network probe node that supports Internet Control Message Protocol echo response. The communication module sends network layer probe packets to the IMS network connectivity probe address through the physical network card according to the set probe period. The set probe period ranges from 1s to 5s, and the network layer probe packet is presented as an Internet Control Message Protocol echo request packet with a system timestamp and an incrementing sequence number.

[0074] The communication module synchronously starts an internal high-precision hardware timer to perform countdown monitoring, and the timeout threshold for countdown monitoring is configured to be between 200ms and 500ms. Then, the communication module listens to the network interface port to capture the response data frames returned by the IP Multimedia Subsystem network. The communication module evaluates the network connectivity based on the return status of the network layer probe packets. The specific round-trip delay calculation and packet loss statistics for network connectivity can be processed according to existing network performance testing techniques. The communication module continuously writes the evaluated network connectivity parameters into the status monitoring register for retrieval by the main control thread.

[0075] S52, the communication module continuously reads network connectivity parameters from the status monitoring register and performs normalization processing on the round-trip delay and packet loss rate parameters according to the pre-calibrated upper limits of round-trip delay and packet loss rate, respectively. The upper limit of round-trip delay ranges from 500ms to 1000ms, and the upper limit of packet loss rate ranges from 10% to 20%. Then, the specific normalization processing is calculated according to the existing extreme value standardization technique, and the lower limit of extreme value standardization is preset to 0, thereby converting the round-trip delay and packet loss rate parameters into dimensionless evaluation parameters within a unified numerical range.

[0076] The communication module calculates a network connectivity assessment value using normalized round-trip delay and packet loss rate parameters, and compares the network connectivity assessment value with a preset communication blocking characteristic threshold. The specific formula for calculating the network connectivity assessment value is as follows: ; In the formula, This is a network connectivity evaluation value; Delay weighting; For round-trip time delay parameters; Weighting for packet loss; This is a parameter for packet loss rate.

[0077] Both the delay weight and the packet loss weight are non-negative normalized constants determined through calibration tests, and their sum is always 1. The delay weight ranges from 0.6 to 0.8, the packet loss weight ranges from 0.2 to 0.4, and the communication blocking feature threshold ranges from 0.75 to 0.90. When the network connectivity evaluation value is greater than the communication blocking feature threshold, the communication module determines that the current network has communication blocking features and generates a blocking signal. The blocking signal is a hardware level pulse triggered by a high level, and the communication module injects the blocking signal into the external interrupt request pin of the main processor, thereby triggering the underlying communication link switching interrupt.

[0078] S53, during the execution of the exception handling service routine for the interruption of the underlying communication link, the communication module reads the basic fusion data sequence, equipment asset code, numerical deviation of each feature parameter, and fault level identifier corresponding to the current alarm event. The communication module reads the exception parameter type from the basic fusion data sequence and determines the deviation interval code corresponding to the numerical deviation based on the preset deviation interval boundary. The value range of the preset deviation interval boundary is 1.5 to 3.0. The communication module combines the equipment asset code, exception parameter type, deviation interval code, and fault level identifier to form an alarm feature index.

[0079] The pre-set alarm short code hash mapping table uses a limited number of alarm feature indices as retrieval keys, with a value range of 256 to 1024, and uses the corresponding mapped short codes as mapping values. The communication module queries the corresponding mapped short code from the pre-set alarm short code hash mapping table based on the alarm feature index, and then writes the mapped short code into the transmit buffer register of the external satellite communication hardware through the serial peripheral interface, thereby forming an alarm data payload that meets the satellite channel bandwidth limit. The remote service takeover platform maintains the same hash mapping relationship as the communication module, and recovers the equipment asset code, abnormal parameter type, deviation range code, and fault level identifier based on the received mapped short code.

[0080] S54, the external satellite communication hardware is pre-configured with satellite network access identifier, operating frequency band, channel parameters, modulation and coding parameters, and the service address corresponding to the remote service takeover platform. The frequency range of the operating frequency band is configured to be 1.5GHz to 2.5GHz, the bandwidth of the channel parameters is configured to be 2MHz to 5MHz, and the forward error correction code rate of the modulation and coding parameters is configured to be 1 / 2 or 3 / 4. After detecting that there is a short code to be transmitted in the transmit buffer register, the external satellite communication hardware activates the radio frequency transceiver link and completes link synchronization and channel access according to the pre-configured satellite network access parameters.

[0081] The external satellite communication hardware uses phase shift keying modulation to perform physical carrier modulation processing on the mapped short code, and sends the radio signal carrying the mapped short code to the low Earth orbit communication satellite node. The low Earth orbit communication satellite node forwards the received radio signal to the corresponding ground gateway. The ground gateway completes signal demodulation and data recovery, and sends the mapped short code to the remote service takeover platform through the pre-configured service bearer link.

[0082] The independent and continuous transmission of microwave electromagnetic signals in the space physical channel traverses severely damaged public network infrastructure, thereby completing a closed loop of the entire data processing chain from multi-source data acquisition and intelligent diagnosis and classification to multi-mode communication scheduling.

[0083] To aid in understanding the technical solution of this invention, the following provides an application example of intelligent early warning communication for power plants based on IMS.

[0084] The application example receives the underlying data stream output by multi-source heterogeneous sensing devices under complex operating conditions in power plants, and pre-configures a network communication resource configuration file containing a historical fault feature matrix and a mapping relationship between multi-level responsible objects, as a benchmark for intelligent diagnosis and classification, multi-mode communication scheduling closed-loop feedback and effect comparison.

[0085] The acquisition module obtains discrete structured parameters output by the underlying physical sensors and continuous image data frames extracted by the video surveillance equipment. Based on the network time protocol service interface and static physical topology configuration file of the underlying system, it assigns absolute system timestamps and three-dimensional physical coordinates to the data samples. Next, the acquisition module performs data parsing for different underlying industrial communication protocols, completes binary packaging and encapsulation by combining the set bit width alignment and byte padding rules, and simultaneously performs access lock restriction operations in conjunction with the memory cross-process communication mechanism. Finally, the storage module receives the processed message, performs address matching operation to obtain the globally unique device asset code associated with the underlying layer, and dynamically allocates a logically contiguous storage area in the system memory heap area, thereby establishing a basic fusion data sequence hash mapping logic with the device asset code as the retrieval key value.

[0086] The diagnostic module reads the basic fusion data sequence based on the set diagnostic cycle, performs matrix comparison calculations on the extracted feature values ​​and preset operating benchmark indicators, extracts discrete deviation differences and combines them to integrate the underlying data to generate a deviation feature vector containing the objective severity of the real-time operating conditions. Then, the diagnostic module performs pattern matching operations with the preset historical fault sample center vectors according to the judgment weight coefficients of each feature parameter in the weighted Euclidean distance calculation formula, and extracts the calculation result with the smallest value to determine the target matching distance. Within this operating cycle, the diagnostic module outputs the fault level identifier associated with the corresponding fault category based on the comparison result of the target matching distance and the preset matching acceptance threshold, and writes it into the first-in-first-out message queue according to the data structure of the same alarm event.

[0087] The control module reads feature data from the message queue in a first-in-first-out order, matches the pin level control parameters with the extracted fault level identifier, and outputs a square wave pulse signal to drive the local field alarm device to perform audible and visual flashing and sounding actions. Simultaneously, the control module extracts the over-limit time threshold corresponding to the fault level identifier in the timeout mapping table and injects it into the overload register of the hardware timer, thereby dynamically starting a step countdown process based on the system clock source. During this countdown cycle, the control module uses a non-blocking background listening thread to continuously monitor the interface status register of the human-machine interaction terminal, and generates a hardware interrupt request when it determines that no response level pulse has been captured and the countdown has reached zero. The hardware interrupt request is used to suspend the local waiting process, and completes the extraction and assembly of the control transfer instructions from the local confirmation stage to the remote communication scheduling stage.

[0088] The communication module receives the control handover instruction, calls the internal audio conversion program to synthesize waveform audio media streams based on the extracted structured parameters, and constructs a session initiation protocol data packet with inserted custom extended fields. It then initiates signaling queuing addressing to the multimedia subsystem network environment to establish an audio and video transmission channel. Next, the communication module initiates network connectivity probing in parallel at the underlying protocol stack, and performs extreme value normalization and weighted evaluation by capturing the round-trip delay and packet loss rate parameters of the network layer probe packets. When the network connectivity assessment value meets the blocking characteristic conditions, the communication module updates the underlying communication link switching status flag from false to true and generates a hardware blocking signal to inject into the external interrupt request pin, forming an emergency intervention trigger mechanism in a non-public network environment. Finally, based on the alarm feature index generated by the combination, the communication module retrieves the corresponding mapped short code through the preset alarm short code hash mapping table, loads it across processes into the transmit buffer register of the external satellite communication hardware, and performs physical carrier modulation and space channel delivery through the radio frequency transceiver link to realize the remote recovery and action guidance closed loop of multi-source early warning scheduling data under extreme network conditions.

[0089] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations can be made to the equipment structure, process flow, step sequence, operating parameters, or working medium type in these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the core technical solutions disclosed above and their equivalent substitutions.

Claims

1. A smart early warning communication method for power plants based on IMS, characterized in that, The method includes the following steps: Obtain discrete structured parameters and continuous streaming media sequences, assign system timestamps and three-dimensional physical coordinates to the discrete structured parameters and the continuous streaming media sequences, and combine and output basic fusion data sequences to establish memory mapping logic between the basic fusion data sequences and device asset codes; Feature values ​​are extracted from the basic fusion data sequence, the deviation of the feature values ​​from the preset operating benchmark indicators is calculated, and the deviation is used to perform pattern matching operation with the preset historical fault feature matrix to generate a fault level identifier. Based on the fault level identifier, output local environmental sound and light control instructions and activate hardware timers. If the hardware timer reaches the set over-limit time threshold and no response signal from the human-machine interaction terminal is captured, generate a control transfer instruction. The system receives the control transfer instruction to extract the target communication address, constructs a session initiation protocol data packet with a custom extended field containing the fault level identifier inserted in the packet header, and sends the session initiation protocol data packet to the IP multimedia subsystem network to establish an audio and video transmission channel. The network layer probe message is sent to the IP Multimedia Subsystem network to assess network connectivity. When communication blocking characteristics are detected, the underlying communication link switching interruption is triggered. The mapping short code corresponding to the basic fusion data sequence and the fault level identifier is extracted and sent using external satellite communication hardware.

2. The intelligent early warning communication method for power plants based on IMS according to claim 1, characterized in that, The steps of obtaining discrete structured parameters and continuous streaming media sequences, assigning system timestamps and three-dimensional physical coordinates to the discrete structured parameters and the continuous streaming media sequences, combining and outputting a basic fused data sequence, and establishing memory mapping logic between the basic fused data sequence and device asset codes include: The discrete structured parameters are divided according to a set sampling period to form a discrete data payload, and the continuous image data of the continuous streaming media sequence is divided according to a set time window to form a streaming media data payload. Configure a unified data header containing the underlying hardware media access control address for the discrete data payload and the streaming media data payload, and combine them to generate the basic fused data sequence; The device asset code is obtained by performing an address matching operation using the three-dimensional physical coordinates and the underlying hardware media access control address. Construct a memory address hash mapping table with the device asset code as the retrieval key and the starting logical address of the basic fusion data sequence storage area as the mapping value.

3. The intelligent early warning communication method for power plants based on IMS according to claim 1, characterized in that, The steps of extracting feature values ​​from the basic fused data sequence and calculating the deviation of the feature values ​​from the preset operating benchmark index include: Extract the feature values ​​from the basic fused data sequence to construct a one-dimensional feature vector; Read the preset operating benchmark index, which includes theoretical expected values ​​and tolerance range parameters; Calculate the difference between the feature value and the corresponding theoretical expected value, and determine the ratio of the difference to the corresponding tolerance range parameter as the value deviation. Combine these to generate a deviation feature vector.

4. The intelligent early warning communication method for power plants based on IMS according to claim 3, characterized in that, The steps for generating a fault level identifier by performing pattern matching operations using the numerical deviation and a preset historical fault feature matrix include: Calculate the weighted Euclidean distance between the deviation feature vector and the fault center vector in the historical fault feature matrix; The smallest calculated result from the weighted Euclidean distances is determined as the target matching distance, and the fault type corresponding to the target matching distance is recorded as the target fault type. When the target matching distance is less than or equal to the matching acceptance threshold, the fault level code pre-associated with the target fault type is read to generate the fault level identifier.

5. The intelligent early warning communication method for power plants based on IMS according to claim 1, characterized in that, The step of generating a control transfer instruction when the hardware timer reaches a set over-time threshold and no response signal from the human-machine interface terminal is captured includes: The over-limit time threshold is obtained based on the fault level identifier and injected into the hardware timer. The stepping operation of the internal register is executed using the system clock source. Start a non-blocking background listening thread to continuously read the interface status register of the human-computer interaction terminal to monitor the action feedback status; When the remaining time of the hardware timer reaches zero and no valid low-level pulse corresponding to the human-machine interaction terminal response signal is captured, the device asset code, the numerical deviation, and the fault level identifier are extracted to generate the control transfer instruction.

6. The intelligent early warning communication method for power plants based on IMS according to claim 1, characterized in that, The steps of receiving the control transfer instruction to extract the target communication address and constructing a session initiation protocol data packet with a custom extended field containing the fault level identifier inserted in the packet header include: The target communication address is extracted by querying the responsibility object mapping table based on the fault level identifier; The numerical deviation and the equipment asset are encoded into a waveform audio file; Write session description protocol data containing the media transmission port into the message body of the session initiation protocol data packet, and insert the custom extended field containing the fault level identifier into the header of the session initiation protocol data packet.

7. The intelligent early warning communication method for power plants based on IMS according to claim 6, characterized in that, The step of sending the Session Initiation Protocol (SIP) data packet to the IP Multimedia Subsystem (IP Subsystem) network to establish an audio / video transmission channel includes: Based on the fault level identifier, the IMS signaling priority parameter is determined, and the session initiation protocol data packet is sent using the IMS signaling priority parameter to perform signaling queuing and routing forwarding; After the session confirmation is completed, the waveform audio file and the corresponding video data segment in the continuous streaming media sequence are respectively encapsulated into an audio real-time transmission protocol data stream and a video real-time transmission protocol data stream. The audio real-time transmission protocol data stream and the video real-time transmission protocol data stream are sent respectively according to the agreed media transmission port to establish the audio and video transmission channel.

8. The intelligent early warning communication method for power plants based on IMS according to claim 1, characterized in that, The steps of sending network layer probe messages to the IP Multimedia Subsystem network to assess network connectivity and triggering a lower-level communication link handover interruption when communication blocking characteristics are detected include: Send the network layer probe message to the IMS network connectivity probe address and capture the response data frame to obtain round-trip delay parameters and packet loss rate parameters; The round-trip delay parameter and the packet loss rate parameter are normalized according to the calibrated upper limit value, respectively. The network connectivity evaluation value is calculated using the normalized parameters. When the network connectivity evaluation value is greater than the communication blocking characteristic threshold, a blocking signal is generated and injected into the external interrupt request pin to trigger the underlying communication link switching interruption.

9. The intelligent early warning communication method for power plants based on IMS according to claim 1, characterized in that, The steps for extracting the mapping short code corresponding to the basic fusion data sequence and the fault level identifier include: During the execution of the exception handling service routine, the exception parameter type is read from the basic fusion data sequence, and the deviation interval code corresponding to the numerical deviation is determined according to the preset deviation interval boundary. The equipment asset code, the abnormal parameter type, the deviation range code, and the fault level identifier are combined to form an alarm feature index; The corresponding mapping short code is extracted from the preset alarm short code hash mapping table based on the alarm feature index.

10. The intelligent early warning communication method for power plants based on IMS according to claim 9, characterized in that, The step of transmitting the mapped short code using external satellite communication hardware includes: Write the mapped short code into the transmit buffer register of the external satellite communication hardware; After detecting that the mapped short code to be transmitted exists in the transmit buffer register, the radio frequency transceiver link is activated and link synchronization and channel access are completed; The mapped short code is processed by physical carrier modulation using phase shift keying modulation, and the wireless signal is sent to a low Earth orbit communication satellite node for forwarding to a remote service takeover platform.

Images