An Automated Troubleshooting and Diagnostic Method for Vacuum Furnaces Based on Communication Networks

Abstract

This invention discloses an automated troubleshooting and diagnosis method for vacuum furnaces based on a communication network, specifically relating to the field of vacuum furnace leak diagnosis technology. It addresses the problem that existing vacuum furnace leak troubleshooting methods require time-consuming preprocessing steps during prolonged equipment downtime to overcome material venting interference, which severely restricts production efficiency. The method acquires real-time readings from furnace pressure sensors, pipeline pressure sensors, and temperature sensors via a communication network and remotely controls the vacuum pump and valves. After evacuating to a set pressure value and closing the main extraction valve, a cross-recursive graph is constructed using furnace pressure and temperature values ​​within a preset time period, and recursive feature parameters are extracted. Simultaneously, adjacent difference signed run distribution analysis is performed on the furnace pressure value sequence to extract tailing feature parameters. These parameters are combined to determine the furnace leak status. Then, using the furnace leak status result as a reference, bypass pipelines are connected one by one for repeated testing, and status changes are compared to locate the leaking pipeline.

Description

Technical Field

[0001] This invention relates to the field of vacuum furnace leak diagnosis technology, and more specifically, to an automated troubleshooting and diagnosis method for vacuum furnaces based on a communication network. Background Technology

[0002] Leaks in vacuum heat treatment furnaces can severely damage equipment operation, product quality, and the lifespan of furnace components. Industrial vacuum furnaces are large in size, with numerous connected valves, instrument joints, and bypass pipelines, resulting in a wide distribution of potential leak points to be inspected. Current technologies using specialized instruments such as helium mass spectrometers for leak detection require connecting the instruments after furnace shutdown and spraying leak-indicating substances point by point, waiting for each instrument to respond, making a complete inspection of a single unit extremely time-consuming. Another method using the furnace's own vacuum instruments involves prolonged vacuuming to fully desorb gases adsorbed on the furnace chamber and pipeline walls, eliminating interference from material venting, and then closing the main valve to measure the static pressure rise. Because the desorption process is slow, this method also requires the vacuum pump to operate continuously for an extended period before testing can begin. Both of these methods require prolonged idle periods when the equipment is not in production, which encroaches on production time and makes it difficult to utilize production breaks flexibly, leading to a decrease in overall equipment operating efficiency.

[0003] Existing methods for detecting leaks in vacuum furnaces require time-consuming pre-processing steps to overcome the interference of material venting from the furnace on leak detection during prolonged equipment downtime in order to obtain effective diagnostic data. The entire process is highly coupled with the equipment's production tasks, which severely restricts production efficiency in diagnostic activities. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides an automated troubleshooting and diagnosis method for vacuum furnaces based on communication networks to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: An automated troubleshooting and diagnostic method for vacuum furnaces based on a communication network includes the following steps: S1. Obtain real-time readings from the furnace pressure sensor, pipeline pressure sensor, and temperature sensor in the vacuum furnace via the communication network; S2. The vacuum pump is started to evacuate the furnace through the communication network. When the furnace pressure sensor reading reaches the set pressure value, the main evacuation valve is closed. S3. During the preset time period after the main exhaust valve is closed, the furnace pressure and furnace temperature values ​​are intermittently recorded based on real-time readings. A cross-recursion graph of pressure and temperature is constructed, and the recursive characteristic parameters of the cross-recursion graph are extracted. S4. Within a preset time period, perform adjacent difference analysis on the furnace pressure value sequence and extract the symbol sequence. Statistically analyze the run length distribution of consecutive identical symbols in the symbol sequence and extract the tailing feature parameters of the run length distribution. S5. Determine the furnace leakage status based on the recursive characteristic parameters and the tailing characteristic parameters, and obtain the furnace leakage status result. S6. Using the furnace leakage status as a reference, the valves connecting the furnace and the bypass pipeline to be tested are opened sequentially through the communication network. After each opening, S2 to S5 are repeated to obtain the leakage status of the bypass pipeline to be tested.

[0006] Furthermore, S1 includes: Real-time readings of the furnace pressure sensor in the vacuum furnace are obtained through a communication network; Real-time readings of the pressure sensors in the pipes of the vacuum furnace are obtained through a communication network. Real-time readings from temperature sensors in the vacuum furnace are obtained via a communication network. The real-time readings of the furnace pressure sensor, the pipeline pressure sensor, and the temperature sensor are synchronously transmitted to the host computer.

[0007] Furthermore, S2 includes: Control commands are transmitted to the host computer via the communication network, and the host computer controls the vacuum pump of the vacuum furnace to start evacuating the furnace chamber via the communication network. Real-time readings from the furnace pressure sensor are continuously acquired during the vacuuming process; The real-time readings of the furnace pressure sensor are continuously compared with the preset pressure values ​​stored in the host computer. When the real-time reading of the furnace pressure sensor reaches the set pressure value, the host computer controls the main exhaust valve to close via the communication network.

[0008] Furthermore, S3 includes: Starting from the moment the main exhaust valve is closed, a preset time period is calculated. Within the preset time period, the real-time readings of the furnace pressure sensor are recorded intermittently at preset time intervals to form a furnace pressure value sequence. Real-time readings of the temperature sensor are recorded intermittently at the same preset time interval within a preset period to form a sequence of furnace temperature values; Using the furnace pressure and temperature sequences as bivariate inputs, a cross-recursive graph is constructed with furnace pressure as the horizontal axis variable and furnace temperature as the vertical axis variable. Recursive feature parameters are extracted from the cross recursion graph. These recursive feature parameters include the recursion rate and the determination rate.

[0009] Furthermore, a cross-recursive graph is constructed with furnace pressure as the horizontal axis variable and furnace temperature as the vertical axis variable, including: The phase space of the furnace pressure value sequence is reconstructed to obtain the pressure phase space vector of the furnace pressure value sequence under the embedding dimension; The phase space of the furnace temperature value sequence is reconstructed to obtain the temperature phase space vector of the furnace temperature value sequence under the same embedding dimension. Calculate the distance between each vector in the pressure phase space vector and each vector in the temperature phase space vector; The calculated distance is compared with a pre-set recursion threshold. When the distance is less than the recursion threshold, the corresponding coordinate position in the cross recursion graph is marked as a recursion point. When the distance is greater than or equal to the recursion threshold, the corresponding coordinate position in the cross recursion graph is marked as a non-recursion point. This forms a two-dimensional binary cross recursive graph consisting of recursive points and non-recursive points.

[0010] Furthermore, S4 includes: The adjacent difference calculation is performed on the furnace pressure value sequence formed within the preset time period to obtain the difference sequence; Extract the sign of each element in the difference sequence, assign the positive difference to the first sign, the negative difference to the second sign, and the zero difference to the third sign, and arrange them in chronological order to form a sign sequence; The run length distribution is obtained by calculating the run lengths of each run consisting of consecutive identical symbols in a statistical symbol sequence. Tail feature parameters that characterize the tail shape of the run length distribution are extracted. These tail feature parameters include the maximum run length and the Pareto distribution shape parameter.

[0011] Furthermore, the symbols of each element in the difference sequence are extracted, including: marking the elements with positive values ​​as the first symbol, the elements with negative values ​​as the second symbol, and the elements with zero values ​​as the third symbol, and arranging the marked symbols in the original time order of the difference sequence to form a symbol sequence.

[0012] Furthermore, tail-like feature parameters characterizing the tail morphology of the run-length distribution are extracted, including: Arrange the run length values ​​in the run length distribution in ascending order to obtain an ordered sequence of run lengths; Take the cumulative distribution function for the ordered sequence of run lengths and plot the cumulative distribution curve of the run length values ​​in a double logarithmic coordinate system; Take the data points in the tail region of the cumulative distribution curve. The tail region is the interval where the run length value is greater than the median run length value. Pareto distribution fitting is performed on the data points in the tail region, and the fitting formula is P(L≥l)=(lmin / l). α, where l is the run length value, lmin is the minimum run length value in the tail region, α is the shape parameter of the Pareto distribution, and P(L≥l) is the cumulative probability that the run length value is greater than or equal to l; The shape parameter of the Pareto distribution and the maximum run length value in the ordered run length sequence are extracted and used together as the tail feature parameter.

[0013] Furthermore, S5 includes: The recursion rate and determination rate in the recursive feature parameters are compared with the pre-stored coupling reference values, respectively. The maximum run length and Pareto distribution shape parameter in the tail feature parameters are compared with the pre-stored run reference values, respectively. When the maximum run length is less than the lower limit of the run length in the run reference value and the shape parameter of the Pareto distribution is greater than the upper limit of the shape parameter in the run reference value, and the recursion rate and the determination rate do not exceed the coupling reference value, it is determined that there is intermittent leakage or sealing defect in the furnace. When both the recursion rate and the determination rate exceed the coupling reference value, it is determined that there is a continuous leak in the furnace. When neither the recursion rate nor the determination rate exceeds the coupling reference value, and the maximum run length is greater than or equal to the lower limit of the run length and the Pareto distribution shape parameter is less than or equal to the upper limit of the shape parameter, the furnace is considered to be sealed normally.

[0014] Furthermore, S6 includes: The results of the furnace leakage status determination are stored as a reference baseline; The valve connecting the furnace and the first bypass pipeline to be tested is opened by controlling the communication network. The overall execution after the furnace is connected to the first bypass pipeline to be tested is based on S2 to S5 to obtain the first leakage state result. The first leakage state result is compared with the reference standard. When the first leakage state result deteriorates relative to the reference standard, it is determined that the first bypass pipeline under test has a leak. Close the valve connecting the furnace to the first bypass pipeline to be tested, and open the valve connecting the furnace to the second bypass pipeline to be tested through the communication network. Repeat S2 to S5 for the entire furnace after it is connected to the second bypass pipeline to be tested to obtain the second leakage state result. The second leakage status result is compared with the reference standard. When the second leakage status result deteriorates relative to the reference standard, it is determined that there is a leak in the second bypass pipeline to be tested. Repeat the above process for the remaining bypass pipes to be tested until the leakage status of all bypass pipes to be tested is determined.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The beneficial effects of this invention are: 1. Real-time readings from furnace pressure sensors, pipeline pressure sensors, and temperature sensors are automatically acquired via communication networks, and the opening and closing of vacuum pumps and valves are remotely controlled. The entire process is automatically executed by the host computer according to a preset procedure, eliminating the need for operators to manually open or close valves or record data during the diagnostic process. This allows leak detection to be completed automatically during production breaks, freeing it from dependence on prolonged equipment downtime. The diagnostic process is based on furnace pressure and temperature values ​​collected within a preset time period after the main exhaust valve is closed. The recursion rate and certainty rate are extracted from the cross-recursion graph. Simultaneously, the tailing feature analysis is performed on the distribution of adjacent difference sign run lengths of the furnace pressure value sequence. The recursive feature parameter reflects the dynamic coupling strength between pressure and temperature, and the tailing feature parameter reflects the unidirectional continuity of the pressure rise direction. These two parameters characterize the leakage behavior from different dimensions, enabling the diagnosis to directly enter the testing stage after the vacuum pump reaches the set pressure value. This avoids the long desorption pretreatment required in existing methods to eliminate material venting interference, significantly reducing the time required for a single diagnosis.

[0016] 2. The determination of furnace leakage status employs a cross-combination logic of recursive and tailing feature parameters, which can distinguish between three states: intermittent leakage or sealing defects, continuous leakage, and normal sealing, rather than simply providing a binary conclusion of whether a single leak exists. This provides a more precise direction for subsequent leak detection and maintenance. After obtaining the leakage status of the furnace itself, it is used as a reference benchmark. By opening bypass pipeline valves one by one and repeating the same vacuuming and feature extraction determination process, the leaking pipeline is located by comparing whether the leakage status deteriorates before and after pipeline connection. This achieves zoned diagnosis of the furnace and each bypass pipeline, avoiding misjudgment of furnace leakage due to pipeline leakage, and eliminating the cumbersome operation of connecting instruments to each pipeline section individually. The entire diagnostic process only involves hardware modifications to the vacuum furnace, including communication and acquisition of sensor data and remote control access to valves and pumps. It does not change the furnace structure or the original operating logic, making it easy to implement and promote on existing industrial vacuum furnaces. Attached Figure Description

[0017] Figure 1 This is a flowchart of an automated troubleshooting and diagnostic method for vacuum furnaces based on a communication network, according to the present invention. Detailed Implementation

[0018] The technical solutions of 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example: Figure 1 This invention provides an automated troubleshooting and diagnostic method for vacuum furnaces based on a communication network, which includes the following steps: S1. Obtain real-time readings from the furnace pressure sensor, pipeline pressure sensor, and temperature sensor in the vacuum furnace via the communication network; S2. The vacuum pump is started to evacuate the furnace through the communication network. When the furnace pressure sensor reading reaches the set pressure value, the main evacuation valve is closed. S3. During the preset time period after the main exhaust valve is closed, the furnace pressure and furnace temperature values ​​are intermittently recorded based on real-time readings. A cross-recursion graph of pressure and temperature is constructed, and the recursive characteristic parameters of the cross-recursion graph are extracted. S4. Within a preset time period, perform adjacent difference analysis on the furnace pressure value sequence and extract the symbol sequence. Statistically analyze the run length distribution of consecutive identical symbols in the symbol sequence and extract the tailing feature parameters of the run length distribution. S5. Determine the furnace leakage status based on the recursive characteristic parameters and the tailing characteristic parameters, and obtain the furnace leakage status result. S6. Using the furnace leakage status as a reference, the valves connecting the furnace and the bypass pipeline to be tested are opened sequentially through the communication network. After each opening, S2 to S5 are repeated to obtain the leakage status of the bypass pipeline to be tested.

[0020] In the specific implementation of S1, real-time readings from a furnace pressure sensor in the vacuum furnace are acquired via a communication network. The furnace pressure sensor is installed on the furnace wall, and its measuring end is directly connected to the internal space of the furnace. The furnace pressure sensor converts the gas pressure inside the furnace into an electrical signal. The furnace pressure sensor is equipped with either an analog output interface or a digital communication interface. When the furnace pressure sensor is configured with an analog output interface, the analog output interface is connected to the input of an analog signal isolation module via a shielded cable. The output of the analog signal isolation module is connected to the input of an analog-to-digital converter (ADC). The ADC converts the analog voltage signal output by the analog signal isolation module into a digital signal, which is then transmitted to a host computer via a communication network. The ADC and the host computer establish a communication connection via industrial Ethernet or fieldbus, using either Modbus TCP or PROFIBUS protocols. When the furnace pressure sensor is configured with a digital communication interface, the digital communication interface directly establishes a communication connection with the host computer through the communication network. The digital communication interface sends the pressure data collected by the furnace pressure sensor to the host computer in the form of digital signals according to the preset communication protocol. After receiving the real-time reading of the furnace pressure sensor, the host computer stores the real-time reading of the furnace pressure sensor in the host computer's memory and adds the first timestamp.

[0021] Real-time readings from a pipe pressure sensor in the vacuum furnace are acquired via a communication network. The pipe pressure sensor is installed on the bypass pipe of the vacuum furnace, and its measuring end is directly connected to the internal space of the bypass pipe. The sensor is positioned between the valve connecting the bypass pipe to the furnace chamber and the valve connecting the bypass pipe to an external gas source, allowing it to measure the gas pressure inside this closed section of the pipe. The pipe pressure sensor converts the gas pressure into an electrical signal and is equipped with either an analog output interface or a digital communication interface. When configured with an analog output interface, the analog output interface is connected to the input of a second analog signal isolation module via a shielded cable. The output of the second analog signal isolation module is connected to the input of a second analog-to-digital converter (ADC). The ADC converts the analog voltage signal output from the second analog signal isolation module into a digital signal, which is then transmitted to a host computer via a communication network. The second ADC and the host computer establish a communication connection via industrial Ethernet or fieldbus, using either Modbus TCP or PROFIBUS protocols. When the pipeline pressure sensor is configured with a digital communication interface, the digital communication interface directly establishes a communication connection with the host computer through a communication network. The digital communication interface transmits the pressure data collected by the pipeline pressure sensor to the host computer in digital signal form according to a preset communication protocol. After receiving the real-time reading from the pipeline pressure sensor, the host computer stores the real-time reading in its memory and adds a second timestamp.

[0022] Real-time readings from temperature sensors within the vacuum furnace are acquired via a communication network. The temperature sensors are installed inside the furnace chamber, with their measuring ends in contact with the internal space. The temperature sensors convert the gas temperature inside the furnace into an electrical signal. The temperature sensors utilize thermocouples or resistance temperature detectors (RTDs) and are equipped with temperature transmitters. These transmitters convert the thermoelectric potential or resistance change generated by the temperature sensors into standard analog or digital signals. When the temperature transmitter outputs a standard analog signal, its output is connected to the input of a third analog signal isolation module via a shielded cable. The output of the third analog signal isolation module is then connected to the input of a third analog-to-digital converter (ADC). The ADC converts the analog signal output from the third analog signal isolation module into a digital signal, which is then transmitted to a host computer via the communication network. The third ADC communicates with the host computer via industrial Ethernet or fieldbus, using either Modbus TCP or PROFIBUS protocols. When the temperature transmitter is configured with a digital communication interface, the digital communication interface of the temperature transmitter directly establishes a communication connection with the host computer through the communication network. The digital communication interface of the temperature transmitter sends the temperature data collected by the temperature sensor to the host computer in the form of digital signals according to the preset communication protocol. After receiving the real-time reading of the temperature sensor, the host computer stores the real-time reading of the temperature sensor in the host computer's memory and adds a third timestamp.

[0023] The real-time readings from the furnace pressure sensor, pipeline pressure sensor, and temperature sensor are synchronously transmitted to the host computer. This synchronous transmission is achieved as follows: The host computer pre-sets a synchronous acquisition period, which is determined by the furnace volume of the vacuum furnace. A larger furnace volume requires a longer synchronous acquisition period; for example, a synchronous acquisition period of 100 milliseconds is set for a furnace volume of 1 cubic meter, and 500 milliseconds is set for a furnace volume of 10 cubic meters. At the beginning of each synchronous acquisition period, the host computer sends a data request command via the communication network to the analog-to-digital converter (ADC) module or digital communication interface corresponding to the furnace pressure sensor, the second ADC module or digital communication interface corresponding to the pipeline pressure sensor, and the third ADC module or digital communication interface corresponding to the temperature sensor. These three data request commands are sent concurrently at the start of the same synchronous acquisition period, with a transmission time deviation not exceeding one clock cycle of the host computer's clock. Upon receiving a data request command, the analog-to-digital converter (ADC) or digital communication interface corresponding to the furnace pressure sensor returns the most recently acquired real-time reading of the furnace pressure sensor to the host computer via the communication network. Similarly, the second ADC or digital communication interface corresponding to the pipeline pressure sensor returns the most recently acquired real-time reading of the pipeline pressure sensor to the host computer via the communication network. The third ADC or digital communication interface corresponding to the temperature transmitter returns the most recently acquired real-time reading of the temperature sensor to the host computer via the communication network. Real-time readings from the furnace pressure sensor, pipeline pressure sensor, and temperature sensor received by the host computer within the same synchronous acquisition cycle are considered synchronous data at the same moment. The host computer adds the same synchronization timestamp to this set of synchronous data and stores it in its memory. A circular buffer is maintained in the host computer's memory. This buffer stores synchronous data acquired from multiple consecutive synchronous acquisition cycles in chronological order of the timestamps. The storage capacity of the circular buffer is determined based on the preset time period and synchronous acquisition cycle in subsequent steps, ensuring that the circular buffer can store all synchronous data acquired within the preset time period. The diagnostic program running on the host computer reads synchronization data from the circular buffer in timestamp order for use in subsequent steps.

[0024] In the specific implementation of S2, control commands are transmitted to the host computer via the communication network. The host computer then controls the vacuum pump of the vacuum furnace to start evacuating the furnace chamber via the communication network. After acquiring and synchronously transmitting the real-time readings of the furnace pressure sensor, pipeline pressure sensor, and temperature sensor in S1 to the host computer, the diagnostic program running on the host computer reads the synchronous data from the loop buffer. Based on the real-time readings of the furnace pressure sensor, the diagnostic program determines whether the vacuum furnace is currently at atmospheric pressure or in a low vacuum state. Once the diagnostic program confirms that the vacuum furnace is currently at atmospheric pressure and the furnace door is closed, it sends a start control command to the vacuum pump controller via the host computer's communication interface. The host computer's communication interface is connected to the vacuum pump controller's communication interface via industrial Ethernet or fieldbus, using Modbus TCP or Profibus protocols. Upon receiving the start control command, the vacuum pump controller closes the main contactor coil circuit of the vacuum pump, causing the main contactor contacts to engage, energizing and starting the vacuum pump motor to begin evacuating the furnace chamber. If the vacuum furnace is equipped with a cascaded composite pump set, which includes a coarse pump and a fine pump, the diagnostic program first sends a start control command to the coarse pump controller via the host computer's communication interface. The coarse pump starts and evacuates the furnace. When the real-time reading of the furnace pressure sensor drops to the switching pressure value between the coarse and fine pumps, the diagnostic program then sends a start control command to the fine pump controller via the host computer's communication interface. The fine pump starts and continues to evacuate the furnace. The switching pressure value is determined based on the fine pump's allowable start pressure, for example, 500 Pascals. Simultaneously with sending the start control command, the host computer marks the start status as "started" and stores it in its memory. This start status mark is used in subsequent steps to determine the vacuum pump's operating status.

[0025] During the vacuuming process, real-time readings of the furnace pressure sensor are continuously acquired. After the vacuum pump starts, the diagnostic program running on the host computer continuously reads the real-time readings of the furnace pressure sensor with synchronization timestamps from the circulating buffer according to the synchronous acquisition cycle set in S1. The circulating buffer is continuously updated according to the order of the timestamps; newly acquired real-time readings of the furnace pressure sensor are appended to the circulating buffer, and the earliest stored real-time readings are overwritten. Each time the diagnostic program reads a real-time reading of the furnace pressure sensor, it compares the real-time reading with the preset pressure value stored in the host computer. During the vacuuming process, the real-time reading of the furnace pressure sensor gradually decreases from atmospheric pressure, and the rate of decrease depends on the pumping speed of the vacuum pump, the furnace volume, and the amount of gas released from the furnace. The real-time readings of the furnace pressure sensor are temporarily stored in the furnace pressure sensor real-time reading variable in the host computer's memory. Each time a new real-time reading of the furnace pressure sensor is read, the host computer overwrites the old value in the furnace pressure sensor real-time reading variable with the new real-time reading. The diagnostic program performs a comparison operation each time the real-time reading of the furnace pressure sensor is updated.

[0026] The real-time readings of the furnace pressure sensor are continuously compared with the preset pressure value stored in the host computer. The preset pressure value is determined as follows: with the vacuum furnace in a cold and empty state, the vacuum pump is started to evacuate the furnace. When the real-time reading of the furnace pressure sensor drops to the target diagnostic starting pressure, the real-time reading of the furnace pressure sensor at this time is recorded as the set pressure value. The target diagnostic starting pressure is determined based on the furnace volume of the vacuum furnace and the sensitivity requirements of the pressure rise rate test within the subsequent preset time period. The target diagnostic starting pressure is selected within the pressure range that reduces the gas release rate of the material inside the furnace to an acceptable level for leakage measurement interference. For example, the target diagnostic starting pressure is selected as 10 Pascals when the furnace volume is 1 cubic meter, and 5 Pascals when the furnace volume is 10 cubic meters. After the set pressure value is determined, it is input and stored in the non-volatile memory of the host computer through the human-machine interface. Each subsequent diagnostic run reads the set pressure value from the non-volatile memory. The specific comparison process is as follows: the diagnostic program reads the set pressure value from the non-volatile memory and compares it with the set pressure value, using the real-time reading of the furnace pressure sensor temporarily stored in the real-time reading variable. If the real-time reading of the furnace pressure sensor is greater than the set pressure value, the comparison result is "not reached," and the diagnostic program continues to wait for the next synchronous acquisition cycle and repeats the process of reading the real-time reading of the furnace pressure sensor and the comparison operation. If the real-time reading of the furnace pressure sensor is less than or equal to the set pressure value, the comparison result is "reached," and the diagnostic program terminates the continuous comparison operation and proceeds to the next step. The diagnostic program also sets a comparison timeout during the continuous comparison process. The comparison timeout is determined based on the maximum time required for the vacuum furnace to normally reach the set pressure value; for example, the comparison timeout is set to twice the time required for normal vacuuming. If the real-time reading of the furnace pressure sensor fails to reach the set pressure value within the comparison timeout period, the diagnostic program determines that the vacuuming process is abnormal, outputs a vacuuming abnormality prompt message through the human-machine interface of the host computer, and terminates the diagnostic process.

[0027] When the real-time reading of the furnace pressure sensor reaches the set pressure value, the host computer controls the main extraction valve to close via the communication network. After the diagnostic program determines that the real-time reading of the furnace pressure sensor has reached the set pressure value, the diagnostic program immediately sends a closing control command to the main extraction valve controller through the host computer's communication interface. The main extraction valve is installed on the main extraction pipeline between the vacuum pump and the furnace. During vacuuming, the main extraction valve is in the open state, connecting the furnace and the vacuum pump's extraction port. After closing, the main extraction valve isolates the gas passage between the furnace and the vacuum pump. The host computer's communication interface is connected to the main extraction valve controller's communication interface via industrial Ethernet or fieldbus, using Modbus TCP or Profibus protocols. Upon receiving the closing control command, the main extraction valve controller drives the main extraction valve's actuator, which can be a pneumatic or electric actuator. When the main extraction valve is configured with a pneumatic actuator, the main extraction valve controller controls the corresponding solenoid valve to switch the compressed air path. The compressed air drives the piston of the pneumatic actuator, which in turn moves the valve core of the main extraction valve to the closed position. When the main extraction valve is configured with an electric actuator, the main extraction valve controller sends a drive signal to the motor of the electric actuator. The motor of the electric actuator drives the valve core of the main extraction valve to rotate or translate to the closed position. After the main extraction valve is closed, a seal is formed between the valve core and the valve seat, isolating the internal space of the furnace from the vacuum pump. A valve position detection device is installed on the main extraction valve. After detecting that the valve core of the main extraction valve has reached the closed position, the valve position detection device sends a closed signal back to the main extraction valve controller. The main extraction valve controller transmits the closed signal to the host computer through the communication network. After receiving the shut-off signal, the diagnostic program records the moment the main suction valve closes as the shut-off time and stores it in the host computer's memory for use in subsequent steps (S3) to calculate a preset time period from the moment the main suction valve closes. After sending the shut-off control command, the diagnostic program also sends a stop control command for the vacuum pump to the vacuum pump controller via the host computer's communication interface. Upon receiving the stop control command, the vacuum pump controller disconnects the main contactor coil circuit of the vacuum pump, stopping the vacuum pump to prevent damage caused by the vacuum pump continuing to pump air from the closed pipeline after the main suction valve closes.

[0028] In the specific implementation of S3, a preset time period is calculated from the moment the main extraction valve is closed. Within this preset time period, the real-time readings of the furnace pressure sensor are recorded intermittently at preset time intervals to form a furnace pressure value sequence. The moment the main extraction valve is closed has already been recorded by the host computer in S2 and stored in the host computer's memory. The diagnostic program reads the moment the main extraction valve is closed from the host computer's memory and uses the moment the main extraction valve is closed as the starting point for the preset time period. The preset time period is stored in the non-volatile memory of the host computer. The preset time period is determined based on the furnace volume of the vacuum furnace and the gas release characteristics of the internal insulation material and structural components. The larger the furnace volume and the larger the surface area of ​​the internal insulation material and structural components, the longer the preset time period is set. This ensures that a sufficient number of real-time readings from the furnace pressure sensor and temperature sensor can be collected within the preset time period to fully reflect the dynamic process of pressure recovery and temperature change. For example, the preset time period is set to 300 seconds when the furnace volume is 1 cubic meter and 600 seconds when the furnace volume is 10 cubic meters. The preset time interval is also pre-stored in the non-volatile memory of the host computer. The preset time interval is determined based on the length of the preset time period and the number of data points required for subsequent phase space reconstruction. The preset time interval is set so that the number of data points collected within the preset time period is not less than the product of the embedding dimension and the delay time step. For example, if the length of the preset time period is 300 seconds, the preset time interval is set to 1 second, and 300 data points can be collected within the preset time period. If the length of the preset time period is 600 seconds, the preset time interval is set to 2 seconds, and 300 data points can also be collected within the preset time period. The diagnostic program reads the real-time reading of the furnace pressure sensor with a synchronization timestamp from the circular buffer every preset time interval within the preset time period. The read real-time readings of the furnace pressure sensor are appended to the furnace pressure value sequence in chronological order of reading time. The furnace pressure value sequence is a one-dimensional array, and each element of the array stores a real-time reading of a furnace pressure sensor. The array index corresponds to the chronological order of reading time. The diagnostic program continues to perform reading operations until the preset time period ends, and the final furnace pressure value sequence includes the real-time readings of all furnace pressure sensors collected within the preset time period.

[0029] The real-time readings of the temperature sensors are recorded intermittently at the same preset time intervals within a preset time period to form a furnace temperature value sequence. This preset time interval is exactly the same as the preset time interval used to form the furnace pressure value sequence, both read from the same storage location in non-volatile memory. The diagnostic program reads the real-time readings of the temperature sensors with synchronization timestamps from the circular buffer every preset time interval within the preset time period. The time at which the real-time readings of the temperature sensors are read is exactly the same as the time at which the real-time readings of the furnace pressure sensors are read, both triggered based on the same synchronization timestamp. The read real-time readings of the temperature sensors are appended sequentially to the furnace temperature value sequence according to the reading time. The furnace temperature value sequence is also a one-dimensional array, with each element storing a real-time reading of a temperature sensor. The array index corresponds to the reading time sequence. The array length of the furnace temperature value sequence is equal to the array length of the furnace pressure value sequence, and two elements at the same index position in both arrays correspond to the real-time readings of the furnace pressure sensor and the temperature sensor collected at the same time. The diagnostic program continues to perform reading operations until the preset time period ends, and the final furnace temperature value sequence includes the real-time readings of all temperature sensors collected within the preset time period.

[0030] Using the furnace pressure and temperature sequences as bivariate inputs, a cross-recursive graph is constructed with furnace pressure as the horizontal axis and furnace temperature as the vertical axis. The construction of the cross-recursive graph first requires phase space reconstruction of the furnace pressure sequence to obtain the pressure phase space vector under the embedding dimension. Phase space reconstruction is achieved using a time-delayed embedding method, which requires determining two parameters: the embedding dimension and the delay time. The embedding dimension is determined based on the length of the furnace pressure sequence using a pseudo-nearest neighbor method. The pseudo-nearest neighbor method is calculated as follows: starting with an embedding dimension of 1, the embedding dimension is gradually increased. At each embedding dimension, the phase space of the furnace pressure sequence is reconstructed, and the change in distance between nearest neighbor pairs in the phase space after increasing the embedding dimension by 1 is calculated. When the ratio of the number of nearest neighbor pairs whose distance change exceeds a preset ratio to the total number of nearest neighbor pairs decreases below a preset ratio, the corresponding embedding dimension is the selected embedding dimension. For example, the preset ratio is set to 5%. The delay time is determined based on the autocorrelation function of the furnace pressure value sequence. The curve of the autocorrelation function value changing with the delay step is calculated. The delay step corresponding to the point where the autocorrelation function value first drops to a multiple of the difference between 1 and the reciprocal of the natural constant e (the maximum value of the autocorrelation function), is then multiplied by a preset time interval to obtain the delay time. After determining the embedding dimension and delay time, the furnace pressure value sequence is reconstructed into a pressure phase space vector. The reconstruction method is as follows: data points of the embedding dimension separated by one delay time in the furnace pressure value sequence are sequentially taken to form a pressure phase space vector. The total number of real-time readings of the furnace pressure sensor included in the furnace pressure value sequence is denoted as the sequence length, the embedding dimension as the embedding dimension, and the delay time step number as the delay time step number, which is obtained by dividing the delay time by the preset time interval. The construction process of the pressure phase space vector is as follows: Starting from the first pressure phase space vector, the nth pressure phase space vector is composed of the real-time readings of the furnace pressure sensor at the nth index, the nth index plus the delay time step, the nth index plus twice the delay time step, and so on, up to the nth index plus the difference of the embedding dimension minus 1 multiplied by the delay time step. The nth value increases from 1 to the difference of the sequence length minus the embedding dimension minus 1 multiplied by the delay time step. Thus, a total of n pressure phase space vectors are obtained, which are the difference of the sequence length minus the embedding dimension minus 1 multiplied by the delay time step. The dimension of each pressure phase space vector is equal to the embedding dimension.

[0031] The furnace temperature value sequence is reconstructed in phase space under the same embedding dimension to obtain a temperature phase space vector. Using the same embedding dimension and the same number of delay time steps as the furnace pressure value sequence, the furnace temperature value sequence is reconstructed in phase space in the same way as the furnace pressure value sequence: a temperature phase space vector is formed by sequentially taking data points of the embedding dimension that are separated by one delay time in the furnace temperature value sequence. The nth temperature phase space vector is composed of the real-time readings of temperature sensors in the furnace temperature value sequence, from the nth index plus the nth index with a delay time step, to the nth index plus twice the delay time step, up to the nth index plus the difference of the embedding dimension minus 1 multiplied by the delay time step. The nth index increases from 1 to the difference of the sequence length minus the embedding dimension minus 1 multiplied by the delay time step, resulting in a total of n temperature phase space vectors. The dimension of each temperature phase space vector is also equal to the embedding dimension. The number of pressure phase space vectors is equal to the number of temperature phase space vectors.

[0032] Calculate the distance between each vector in the pressure phase space vector and each vector in the temperature phase space vector. The distance is calculated using Euclidean distance. For a certain pressure phase space vector and a certain temperature phase space vector, calculate the sum of squares of the differences between the corresponding coordinate components of the pressure phase space vector and the temperature phase space vector, and then take the square root of the sum of squares to obtain the Euclidean distance between the pressure phase space vector and the temperature phase space vector.

[0033] The calculated distance is compared with a pre-set recursion threshold. When the distance is less than the recursion threshold, the corresponding coordinate position in the cross-recursion graph is marked as a recursive point; when the distance is greater than or equal to the recursion threshold, the corresponding coordinate position in the cross-recursion graph is marked as a non-recursive point. The recursion threshold is pre-stored in the non-volatile memory of the host computer. The recursion threshold is determined as follows: take the set of distances between all pairs of pressure phase space vectors and all temperature phase space vectors, calculate the statistical distribution of this set of distances, and take the distance value at the preset quantile of this statistical distribution as the recursion threshold. The preset quantile is determined based on the measurement noise levels of the furnace pressure sensor and the temperature sensor. The higher the measurement noise level, the higher the preset quantile is set to reduce false recursion points caused by measurement noise. For example, when the measurement noise of the furnace pressure sensor is 0.1% of full scale and the measurement noise of the temperature sensor is 0.1% of full scale, the preset quantile is 10%, and the recursion threshold is the 10th quantile of the set of distances between all pairs of pressure phase space vectors and temperature phase space vectors. The cross-recursion graph is a two-dimensional square matrix. The number of rows in the two-dimensional square matrix equals the number of pressure phase space vectors, and the number of columns equals the number of temperature phase space vectors. A specific row and column position in the two-dimensional square matrix corresponds to the distance comparison result between a specific pressure phase space vector and a specific temperature phase space vector. When the distance between these two vectors is less than the recursion threshold, the specified row and column position in the two-dimensional square matrix is ​​marked as a recursive point and assigned a value of 1. When the distance between these vectors is greater than or equal to the recursion threshold, the specified row and column position in the two-dimensional square matrix is ​​marked as a non-recursive point and assigned a value of 0. After calculating and comparing the distances between all pressure and temperature phase space vectors pairwise, a two-dimensional binary cross-recursion graph is formed, consisting of recursive and non-recursive points.

[0034] Recursive feature parameters are extracted from the cross-recursion graph, including the recursion rate and the determination rate. The recursion rate is defined as the number of recursive points in the two-dimensional binary cross-recursion graph divided by the total number of coordinate points in the graph; that is, the number of points with a value of 1 divided by the total size of the two-dimensional matrix. The determination rate is defined as the number of recursive points in the two-dimensional binary cross-recursion graph that form a straight line segment parallel to the main diagonal divided by the total number of recursive points. The method for determining a straight line segment parallel to the main diagonal is as follows: a line segment search is performed along the main diagonal of the two-dimensional binary cross-recursion graph. A sequence of consecutive recursive points parallel to the main diagonal is identified as a straight line segment. The minimum length threshold of the straight line segment is pre-stored in the non-volatile memory of the host computer. The minimum length threshold is determined based on the embedding dimension; the larger the embedding dimension, the larger the minimum length threshold is set. For example, the minimum length threshold is set to twice the embedding dimension. The number of recursive points on all identified line segments parallel to the main diagonal is summed to obtain the total number of recursive points forming line segments parallel to the main diagonal. This number is then divided by the total number of recursive points to obtain the determination rate. The diagnostic program stores the calculated recursion rate and determination rate in the host computer's memory for subsequent use by S5.

[0035] In the specific implementation of S4, the adjacent difference calculation is performed on the furnace pressure value sequence formed within the preset time period to obtain the difference sequence. The furnace pressure value sequence has already been formed in S3 by intermittently recording the real-time readings of the furnace pressure sensors at preset time intervals within the preset time period. The diagnostic program reads the furnace pressure value sequence from the host computer's memory. The furnace pressure value sequence is a one-dimensional array containing all the real-time readings of the furnace pressure sensors collected within the preset time period, and the array index corresponds to the chronological order of reading time. The diagnostic program successively subtracts the real-time readings of two furnace pressure sensors at adjacent index positions in the furnace pressure value sequence, subtracting the real-time reading of the furnace pressure sensor at the previous index position from the real-time reading of the furnace pressure sensor at the later index position to obtain the adjacent difference value. Specifically, for the real-time readings of two furnace pressure sensors in the furnace pressure value sequence, one indexed at position number and the other at position number plus one, the difference between the real-time reading of the furnace pressure sensor at position number plus one and the real-time reading of the furnace pressure sensor at position number is calculated. This difference is used as the i-th adjacent difference value. The position numbers are sequentially increased from one to the array length of the furnace pressure value sequence minus one, resulting in an array length minus one adjacent difference value. All adjacent difference values ​​are arranged in ascending order of position number to form a difference sequence, which is also a one-dimensional array. The array length of the difference sequence is one less than the array length of the furnace pressure value sequence. The diagnostic program stores the difference sequence in the memory of the host computer.

[0036] The program extracts the sign of each element in the difference sequence, assigning positive differences as the first sign, negative differences as the second sign, and zero differences as the third sign, arranging them in chronological order to form a sign sequence. Each element in the difference sequence has three possible values: positive, negative, or zero. The diagnostic program iterates through each element in the difference sequence, determining the value of each element. When an element's value is greater than zero, the program marks it as the first sign, which can be represented by a numerical code, such as positive one. When an element's value is less than zero, the program marks it as the second sign, such as negative one. When an element's value is zero, the program marks it as the third sign, such as zero. The program then arranges the marked signs in the original chronological order of the corresponding elements in the difference sequence to form a sign sequence. The symbol sequence is a one-dimensional array, with the same length as the difference sequence array. Each element in the symbol sequence can be one of the first, second, or third symbols. The array indices of the symbol sequence and the difference sequence correspond one-to-one, with symbols at the same index corresponding to adjacent difference values ​​at the same index. The diagnostic program stores the symbol sequence in the host computer's memory.

[0037] The run length distribution is obtained by calculating the run lengths of each run consisting of consecutive identical symbols in a sequence of symbols. The diagnostic program traverses the sequence starting from the first element, identifying boundary positions where symbol values ​​change. A continuous interval where symbol values ​​remain unchanged is defined as a run, and the number of symbols within a run is defined as its length. The minimum run length is one; when two adjacent elements have different symbol values, the preceding element constitutes a run of length one. Specifically, the diagnostic program sets the current position pointer to the first element of the sequence, sets the starting position of the current run to the current position pointer, and sets the symbol value of the current run to the symbol value of the element pointed to by the current position pointer. Then, the program moves the current position pointer one position forward and compares the symbol value of the element pointed to by the current position pointer with the symbol value of the current run. If they are the same and the current position pointer has not exceeded the end of the sequence, the program continues to move the current position pointer forward and repeats the comparison process. If the run lengths differ or the current position pointer has exceeded the end of the symbol sequence, the diagnostic program marks a run as finished. The run length is equal to the current position pointer value minus the value of the current run's start position, and this run length is recorded in the run length set. The diagnostic program then updates the current run's start position to the current position pointer and updates the current run's symbol value to the symbol value of the element pointed to by the current position pointer in the symbol sequence, repeating this process until the entire symbol sequence has been traversed. After traversal, the diagnostic program obtains a run length set containing the run length values ​​of all runs in the symbol sequence. All run length values ​​in the run length set are arranged in the order they were extracted to obtain the run length distribution, which is a one-dimensional array. Each element of the array stores the run length value of one run. The diagnostic program stores the run length distribution in the host computer's memory.

[0038] Tail-like features representing the tail shape of the run-length distribution are extracted. These tail-like features include the maximum run-length and the Pareto distribution shape parameter. The diagnostic program reads the run-length distribution from the host computer's memory and arranges the run-length values ​​in ascending order to obtain an ordered run-length sequence. The ordered run-length sequence is a one-dimensional array, where each element is a run-length values ​​from the run-length distribution arranged in ascending order. The diagnostic program then calculates the cumulative distribution function for the ordered run-length sequence. The cumulative distribution function is calculated as follows: for each unique run-length value in the ordered run-length sequence, the number of run-length values ​​less than or equal to that value is counted. This count is then divided by the total number of run-length values ​​in the ordered run-length sequence to obtain the cumulative probability value corresponding to that run-length value. The diagnostic program plots the cumulative distribution curve of run length values ​​in a log-log coordinate system. In this system, the horizontal axis represents run length values ​​(base 10), and the vertical axis represents cumulative probability values ​​(base 10). The program marks each run length value and its corresponding cumulative probability value as a data point in the log-log coordinate system, and connects all data points in ascending order of run length values ​​to form the cumulative distribution curve. The program selects data points in the tail region of the cumulative distribution curve, which represents the interval where run length values ​​are greater than the median run length value. The median run length value is the run length value located in the middle of the ordered run length sequence. When the array length of the ordered run length sequence is odd, the median run length value is taken as the run length value at the index of the array length plus one divided by two. When the array length is even, the median run length value is taken as the arithmetic mean of the run length value at the index of the array length divided by two and the run length value at the index of the array length divided by two plus one. The diagnostic procedure retains only data points in the cumulative distribution curve whose horizontal coordinates are greater than the median of the run length as data points in the tail region.

[0039] Pareto distribution fitting is performed on the data points in the tail region, and the fitting formula is P(L≥l)=(lmin / l). αWhere l is the run length value, lmin is the minimum run length value in the tail region, α is the Pareto distribution shape parameter, and P(L≥l) is the cumulative probability that the run length value is greater than or equal to l. The minimum run length value in the tail region is the run length value with the smallest x-coordinate among the data points in the tail region. The specific process of Pareto distribution fitting is as follows: The diagnostic program performs linear regression on the data points in the tail region in a double logarithmic coordinate system, using the logarithmic value of the x-coordinate of the data points in the tail region in the double logarithmic coordinate system as the independent variable and the logarithmic value of the y-coordinate of the data points in the tail region in the double logarithmic coordinate system as the dependent variable, and fits a straight line using the least squares method. The absolute value of the slope of the straight line is the Pareto distribution shape parameter. The diagnostic program takes the value of the last element of the array from the ordered sequence of run lengths as the maximum run length value. The diagnostic program uses the Pareto distribution shape parameter and the maximum run length value together as tail feature parameters and stores them in the memory of the host computer for subsequent use by S5.

[0040] In the specific implementation of S5, the recursion rate and determination rate in the recursive feature parameters are compared with the pre-stored coupling reference values. The recursion rate and determination rate have been extracted from the cross-recursion graph in S3 and stored in the host computer's memory. The diagnostic program reads the recursion rate and determination rate from the host computer's memory. The coupling reference values ​​are pre-stored in the host computer's non-volatile memory. The coupling reference values ​​include a recursion rate reference threshold and a determination rate reference threshold. The recursion rate reference threshold and determination rate reference threshold are pre-determined and stored in the following way: Under the confirmed condition that the vacuum furnace is in a sealed and normal state, that is, under the condition that the vacuum furnace has been fully leak-tested by a helium mass spectrometer and confirmed to be leak-free, and the metallographic structure and mechanical properties of the products are qualified in multiple consecutive production cycles, S1 to S3 are executed to obtain the cross-recursion graph under the sealed and normal state, and the recursion rate and determination rate are extracted from the cross-recursion graph under the sealed and normal state. Repeat steps S1 to S3 multiple times, for example, 10 times, each time in a cold, empty vacuum furnace state, to obtain multiple recursion rate values ​​and multiple determination rate values ​​under normal sealing conditions. The maximum recursion rate value under these conditions is taken as the recursion rate reference threshold, and the minimum determination rate value is taken as the determination rate reference threshold. Both the recursion rate reference threshold and the determination rate reference threshold are stored as a coupling reference value in the non-volatile memory of the host computer. The diagnostic program compares the recursion rate obtained in S3 with the recursion rate reference threshold to determine if the recursion rate exceeds the threshold. Similarly, the diagnostic program compares the determination rate obtained in S3 with the determination rate reference threshold to determine if the determination rate exceeds the threshold. If both the recursion rate and determination rate exceed the threshold, it is determined that both the recursion rate and determination rate exceed the coupling reference value. If neither the recursion rate nor the determination rate exceeds the threshold, it is determined that neither the recursion rate nor the determination rate exceeds the coupling reference value.

[0041] The maximum run length and Pareto distribution shape parameter in the tailing characteristic parameters are compared with pre-stored run reference values. The maximum run length and Pareto distribution shape parameter have been extracted in S4 and stored in the host computer's memory. The diagnostic program reads the maximum run length and Pareto distribution shape parameter from the host computer's memory. The run reference value is pre-stored in the host computer's non-volatile memory and includes a lower limit for the run length and an upper limit for the shape parameter. The lower limit for the run length and the upper limit for the shape parameter are pre-determined and stored as follows: Under the confirmed condition that the vacuum furnace is in a sealed normal state, S1 to S4 are executed to obtain the run length distribution under the sealed normal state. The maximum run length and Pareto distribution shape parameter are extracted from the run length distribution under the sealed normal state. S1 to S4 are repeated multiple times, for example, 10 times, each time under the cold empty state of the vacuum furnace, to obtain multiple maximum run length values ​​and multiple Pareto distribution shape parameter values ​​under the sealed normal state. The minimum maximum run length value under multiple normal sealing conditions is taken as the lower limit of the run length, and the maximum value of the Pareto distribution shape parameter under multiple normal sealing conditions is taken as the upper limit of the shape parameter. The lower limit of the run length and the upper limit of the shape parameter are stored together as run reference values ​​in the non-volatile memory of the host computer. The diagnostic program compares the maximum run length obtained in S4 with the lower limit of the run length to determine if the maximum run length is less than the lower limit. The diagnostic program compares the Pareto distribution shape parameter obtained in S4 with the upper limit of the shape parameter to determine if the Pareto distribution shape parameter is greater than the upper limit. If the maximum run length is less than the lower limit, the maximum run length is determined not to meet the lower limit condition; if the maximum run length is greater than or equal to the lower limit, the maximum run length is determined to meet the lower limit condition. If the Pareto distribution shape parameter is greater than the upper limit, the Pareto distribution shape parameter is determined not to meet the upper limit condition; if the Pareto distribution shape parameter is less than or equal to the upper limit, the Pareto distribution shape parameter is determined to meet the upper limit condition.

[0042] When the maximum run length is less than the lower limit of the run length reference value, the Pareto distribution shape parameter is greater than the upper limit of the shape parameter reference value, and both the recursion rate and the determination rate do not exceed the coupling reference value, the furnace is determined to have intermittent leakage or sealing defects. The diagnostic program performs a logical AND operation on three conditions: condition one is whether the maximum run length is less than the lower limit of the run length; condition two is whether the Pareto distribution shape parameter is greater than the upper limit of the shape parameter; and condition three is whether both the recursion rate and the determination rate do not exceed the coupling reference value. When all three conditions are met, the diagnostic program generates a furnace leakage status result of intermittent leakage or sealing defects. The physical basis for this judgment is as follows: the maximum run length being less than the lower limit of the run length indicates that the tail of the run length distribution is truncated, and there is a lack of long run length in the direction of pressure rise. The Pareto distribution shape parameter being greater than the upper limit of the shape parameter indicates that the tail of the run length distribution decays sharply. Both of these factors together indicate that the direction of pressure rise is repeatedly interrupted, and the pressure rise process does not have unidirectional continuity. Furthermore, the fact that neither the recursion rate nor the certainty rate exceeds the coupling reference value indicates that the degree of dynamic coupling between pressure and temperature does not exceed the fluctuation range under normal sealing conditions, indicating that there is no large amount of external gas continuously entering the furnace. The contradictory pattern between the interrupted direction and the low degree of coupling is consistent with the physical characteristics of intermittent leakage or fretting of the seal: when the leakage channel is intermittently opened and closed, each opening time is short, and the amount of external gas entering the furnace is insufficient to form a stable coupling between pressure and temperature, but a single gas influx is enough to create a disturbance interruption in the direction of pressure rise.

[0043] When both the recursion rate and the certainty rate exceed the coupling reference value, a persistent leak is determined to exist in the furnace. The diagnostic program performs a logical AND operation on whether the recursion rate exceeds the recursion rate reference threshold and whether the certainty rate exceeds the certainty rate reference threshold. When both conditions are met simultaneously, the diagnostic program generates a furnace leak status result of persistent leak. The physical basis for this determination is that when both the recursion rate and the certainty rate exceed the recursion rate reference threshold, it indicates that a strong dynamic coupling has occurred between pressure and temperature, exceeding the fluctuation range under normal sealing conditions. This strong coupling originates from the continuous inflow of external gas into the furnace. The continuous inflow of external gas simultaneously generates continuous disturbances to the furnace pressure and furnace temperature, significantly enhancing the recursive structure between the pressure phase space vector and the temperature phase space vector. This is manifested as dense recursion points in the cross-recursion graph, forming long straight lines along the main diagonal. The characteristics of persistent leaks are that the leak channel remains open, external gas continuously flows into the furnace, the pressure recovery process is relatively stable, and the direction is unidirectional and continuous. Therefore, the tailing characteristic parameters may still be within the range of the run reference value, but the recursive characteristic parameters will inevitably exceed the coupling reference value.

[0044] When both the recursion rate and the determination rate do not exceed the coupling reference value, and the maximum run length is greater than or equal to the lower limit of the run length and the Pareto distribution shape parameter is less than or equal to the upper limit of the shape parameter, the furnace seal is determined to be normal. The diagnostic program performs a logical AND operation on the conditions that the recursion rate and the determination rate do not exceed the coupling reference value, the condition that the maximum run length is greater than or equal to the lower limit of the run length, and the condition that the Pareto distribution shape parameter is less than or equal to the upper limit of the shape parameter. When all three conditions are met simultaneously, the diagnostic program generates a furnace leakage status result indicating that the seal is normal. The physical basis for this judgment is as follows: The fact that neither the recursion rate nor the certainty rate exceeds the coupling reference value indicates no abnormal dynamic coupling between pressure and temperature, and no indication of external gas ingress; the maximum run length being greater than or equal to the lower limit of the run length indicates a long run in the direction of pressure recovery; and the Pareto distribution shape parameter being less than or equal to the upper limit of the shape parameter indicates a gradual decline at the tail of the run length distribution. Both of these factors together indicate that the pressure recovery process is unidirectionally continuous, and the pressure value maintains a monotonically increasing trend without abnormal interruption within the preset time period, consistent with the characteristic of a stable pressure recovery driven by the slow release of material inside the furnace when the seal is intact. The diagnostic program stores the final furnace leakage status result in the host computer's memory. The furnace leakage status result is one of three values: intermittent leakage or sealing defect, continuous leakage, or normal sealing, which is used as a reference benchmark by S6 in subsequent steps.

[0045] In the specific implementation of S6, the furnace leakage status result obtained is stored as a reference baseline. The furnace leakage status result has already been determined by the diagnostic program in S5 and stored in the host computer's memory. The value of the furnace leakage status result is one of three: intermittent leakage or sealing defect, continuous leakage, or normal sealing. The diagnostic program reads the furnace leakage status result from the host computer's memory, copies the furnace leakage status result to an independent storage location in the host computer's memory, marks it as a reference baseline, and attaches a reference baseline label. The reference baseline label contains the furnace identifier and the determination timestamp. The furnace identifier is the unique identification number of the vacuum furnace being diagnosed, and the determination timestamp is the moment when the diagnostic program completes the S5 determination. The reference baseline remains unchanged in the subsequent leakage status determination process of all bypass pipes under test. Only when a complete round of S1 to S5 diagnostic procedures is re-executed for the entire vacuum furnace and a new furnace leakage status result is generated will the diagnostic program overwrite the old reference baseline with the new furnace leakage status result. The diagnostic program stores the reference benchmark in the non-volatile memory of the host computer to prevent loss in case of accidental power failure. After storage, the reference benchmark value is output through the human-machine interface of the host computer, prompting the operator that the leakage status of the furnace itself has been determined.

[0046] The valve connecting the furnace to the first bypass pipe under test is opened via the communication network. Steps S2 to S5 are executed on the entire furnace connected to the first bypass pipe under test, yielding the first leakage status result. Each bypass pipe of the vacuum furnace is connected to the furnace via a corresponding valve. During the execution of steps S2 to S5, this valve is closed to isolate the bypass pipe from the furnace. The diagnostic program contains a pre-stored list of bypass pipes under test, arranged in a preset order by the pipe number and the communication address of the valve connecting the furnace to the bypass pipe. The diagnostic program reads the number of the first bypass pipe under test and the corresponding communication address of the valve connecting the furnace and the first bypass pipe from the list of bypass pipes under test. It then sends an opening control command to the controller of the valve connecting the furnace and the first bypass pipe under test via the communication interface of the host computer. The host computer's communication interface connects to the controller's communication interface via industrial Ethernet or fieldbus, using Modbus TCP or Profibus protocols. Upon receiving the opening control command, the controller of the valve connecting the furnace and the first bypass pipe under test actuates the valve's actuator, opening the valve and connecting the furnace's internal space with the first bypass pipe's internal space. After the valve opens, the diagnostic program executes steps S2 to S5 for the entire furnace connected to the first bypass pipe under test. During S2, the vacuum is applied to the entire space including the furnace and the first bypass pipe under test. After closing the main exhaust valve, the pressure recovery range is also the entire space including the furnace and the first bypass pipe under test. Since the volume of the first bypass pipe is much smaller than the furnace volume, the added volume has a negligible impact on the test results. However, if there is a leak in the first bypass pipe, external gas will enter the aforementioned space through the leak point, thus altering the dynamic characteristics of the pressure recovery. The processes from S2 to S5 are completely consistent with the furnace self-diagnosis. The preset time period and preset time interval in S3 use the same settings as during furnace self-diagnosis. The processes of constructing the cross-recursive graph and extracting recursive feature parameters in S3 are based on the real-time readings of the furnace pressure sensor and temperature sensor collected after the furnace is connected to the first bypass pipe under test. The operation on the furnace pressure value sequence in S4 is also based on the data collected after the furnace is connected to the first bypass pipe under test. S5 uses the same coupling reference value and run reference value as during furnace self-diagnosis for comparison and judgment.After completing S5, the diagnostic program obtains a leakage status result, which is named the first leakage status result. The value of the first leakage status result is also one of the three: intermittent leakage or sealing defect, continuous leakage, or normal sealing. The diagnostic program stores the first leakage status result in the memory of the host computer.

[0047] The first leakage condition result is compared with a reference baseline. When the first leakage condition result deteriorates relative to the reference baseline, the first bypass pipeline under test is determined to have a leak. Deterioration is defined as the first leakage condition result shifting towards a persistent leakage category compared to the reference baseline category in a ranking of persistent leakage, intermittent leakage or sealing defects, and normal sealing. This ranking is from best to worst: normal sealing is best, intermittent leakage or sealing defects is second best, and persistent leakage is worst. The diagnostic procedure determines the position number of the reference baseline category and the first leakage condition result category from the ranking. The position number indicates the category's ranking; a higher position number indicates a more deteriorated category. The diagnostic program compares the position number of the first leakage state result with the position number of the reference baseline. If the position number of the first leakage state result is greater than the position number of the reference baseline, the program determines that the first leakage state result has deteriorated relative to the reference baseline. In this case, the diagnostic program determines that the first bypass pipe under test has a leak, stores the corresponding number of the first bypass pipe under test in the host computer's memory, and marks it as having a leak. If the position number of the first leakage state result is less than or equal to the position number of the reference baseline, the program determines that the first leakage state result has not deteriorated relative to the reference baseline. In this case, the program determines that the first bypass pipe under test does not have a leak, stores the corresponding number of the first bypass pipe under test in the host computer's memory, and marks it as not having a leak. The physical logic of deterioration is as follows: connecting a healthy bypass pipe should not degrade the overall sealing state of the furnace. The overall leakage state of the furnace after connecting the bypass pipe under test should remain consistent with the furnace's own leakage state. If the overall leakage state of the furnace after connecting the bypass pipe under test deteriorates, the new leakage source must originate from the bypass pipe under test itself.

[0048] The valve connecting the furnace to the first bypass pipeline under test is closed. The valve connecting the furnace to the second bypass pipeline under test is opened via the communication network. Steps S2 to S5 are repeated for the entire furnace connected to the second bypass pipeline under test, yielding the second leakage status result. After the diagnostic program determines the first bypass pipeline under test, it sends a closing control command to the controller of the valve connecting the furnace to the first bypass pipeline under test via the host computer's communication interface. The controller of the valve connecting the furnace to the first bypass pipeline under test closes the valve, isolating the passage between the furnace and the first bypass pipeline under test. The diagnostic program reads the number of the second bypass pipeline under test and the corresponding communication address of the valve connecting the furnace to the second bypass pipeline under test from the list of bypass pipelines under test. It then sends an opening control command to the controller of the valve connecting the furnace to the second bypass pipeline under test via the host computer's communication interface. The valve connecting the furnace to the second bypass pipeline under test opens, connecting the internal space of the furnace to the internal space of the second bypass pipeline under test. The diagnostic procedure executes steps S2 to S5 after the furnace is connected to the second bypass pipe under test. The process is completely consistent with that of testing the first bypass pipe. After completing step S5, the second leakage status result is obtained. The value of the second leakage status result is one of three: intermittent leakage or sealing defect, continuous leakage, or normal sealing. The second leakage status result is compared with the reference benchmark in the same way as the comparison between the first leakage status result and the reference benchmark: when the second leakage status result deteriorates relative to the reference benchmark, it is determined that the second bypass pipe under test has a leak, and the corresponding number of the second bypass pipe under test is stored in the memory of the host computer and marked as having a leak; when the second leakage status result does not deteriorate relative to the reference benchmark, it is determined that the second bypass pipe under test does not have a leak, and the corresponding number of the second bypass pipe under test is stored in the memory of the host computer and marked as having no leak.

[0049] The above process is repeated sequentially for the remaining bypass pipes to be tested until the leakage status of all bypass pipes to be tested is determined. The diagnostic program, following a preset order in the list of bypass pipes to be tested, performs the following steps for each bypass pipe to be tested from the third to the next: closing the valve connecting the previous bypass pipe to the furnace, opening the valve connecting the furnace to the current bypass pipe, repeating steps S2 to S5 for the entire bypass pipe connected to the furnace, comparing the obtained leakage status result with a reference standard, and determining whether the current bypass pipe has a leak. This process continues until all bypass pipes in the list of bypass pipes to be tested have been determined. After all bypass pipes to be tested have been determined, the diagnostic program outputs a summary determination result through the human-machine interface of the host computer. The summary determination result includes the reference standard value, the number of each bypass pipe to be tested, and the determination conclusion of whether each bypass pipe has a leak or not. If a bypass pipeline under test is determined to have a leak, the diagnostic program simultaneously outputs the pipeline's number along with a warning label in the summary judgment results. For example, the pipeline's number may be displayed with a different color or symbol than those of bypass pipelines without leaks, prompting the operator to perform targeted helium mass spectrometry re-inspection or repair on that pipeline. The entire diagnostic process from S1 to S6 runs automatically on the host computer, requiring no manual intervention from the operator in operating the vacuum furnace valves, starting or stopping the vacuum pump, or recording data during the diagnostic process. Operators only need to ensure the vacuum furnace is cold and the furnace door is closed before starting the diagnostic process, and to view the summary judgment results output by the host computer's human-machine interface after the diagnostic is completed.

[0050] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0051] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0052] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0053] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0054] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0055] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0056] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A communication network-based automatic troubleshooting and diagnosis method for a vacuum furnace, characterized in that, Includes the following steps: S1. Obtain real-time readings from the furnace pressure sensor, pipeline pressure sensor, and temperature sensor in the vacuum furnace via the communication network; S2. The vacuum pump is started to evacuate the furnace through the communication network. When the furnace pressure sensor reading reaches the set pressure value, the main evacuation valve is closed. S3. During the preset time period after the main exhaust valve is closed, the furnace pressure and furnace temperature values ​​are intermittently recorded based on real-time readings. A cross-recursion graph of pressure and temperature is constructed, and the recursive characteristic parameters of the cross-recursion graph are extracted. S4. Within a preset time period, perform adjacent difference analysis on the furnace pressure value sequence and extract the symbol sequence. Statistically analyze the run length distribution of consecutive identical symbols in the symbol sequence and extract the tailing feature parameters of the run length distribution. S5. Determine the furnace leakage status based on the recursive characteristic parameters and the tailing characteristic parameters, and obtain the furnace leakage status result. S6. Using the furnace leakage status as a reference, the valves connecting the furnace and the bypass pipeline to be tested are opened sequentially through the communication network. After each opening, S2 to S5 are repeated to obtain the leakage status of the bypass pipeline to be tested.

2. The method according to claim 1, wherein, S1 includes: Real-time readings of the furnace pressure sensor in the vacuum furnace are obtained through a communication network; Real-time readings of the pressure sensors in the pipes of the vacuum furnace are obtained through a communication network. Real-time readings from temperature sensors in the vacuum furnace are obtained via a communication network. The real-time readings of the furnace pressure sensor, the pipeline pressure sensor, and the temperature sensor are synchronously transmitted to the host computer.

3. The method according to claim 1, wherein S2 include: Control commands are transmitted to the host computer via the communication network, and the host computer controls the vacuum pump of the vacuum furnace to start evacuating the furnace chamber via the communication network. Real-time readings from the furnace pressure sensor are continuously acquired during the vacuuming process; The real-time readings of the furnace pressure sensor are continuously compared with the preset pressure values ​​stored in the host computer. When the real-time reading of the furnace pressure sensor reaches the set pressure value, the host computer controls the main exhaust valve to close via the communication network.

4. The automated troubleshooting and diagnostic method for vacuum furnaces based on a communication network according to claim 1, characterized in that, S3 includes: Starting from the moment the main exhaust valve is closed, a preset time period is calculated. Within the preset time period, the real-time readings of the furnace pressure sensor are recorded intermittently at preset time intervals to form a furnace pressure value sequence. Real-time readings of the temperature sensor are recorded intermittently at the same preset time interval within a preset period to form a sequence of furnace temperature values; Using the furnace pressure and temperature sequences as bivariate inputs, a cross-recursive graph is constructed with furnace pressure as the horizontal axis variable and furnace temperature as the vertical axis variable. Recursive feature parameters are extracted from the cross recursion graph. These recursive feature parameters include the recursion rate and the determination rate.

5. The automated troubleshooting and diagnostic method for vacuum furnaces based on a communication network according to claim 4, characterized in that, Construct a cross-recursive graph with furnace pressure as the horizontal axis variable and furnace temperature as the vertical axis variable, including: The phase space of the furnace pressure value sequence is reconstructed to obtain the pressure phase space vector of the furnace pressure value sequence under the embedding dimension; The phase space of the furnace temperature value sequence is reconstructed to obtain the temperature phase space vector of the furnace temperature value sequence under the same embedding dimension. Calculate the distance between each vector in the pressure phase space vector and each vector in the temperature phase space vector; The calculated distance is compared with a pre-set recursion threshold. When the distance is less than the recursion threshold, the corresponding coordinate position in the cross recursion graph is marked as a recursion point. When the distance is greater than or equal to the recursion threshold, the corresponding coordinate position in the cross recursion graph is marked as a non-recursion point. This forms a two-dimensional binary cross recursive graph consisting of recursive points and non-recursive points.

6. The automated troubleshooting and diagnosis method for vacuum furnaces based on a communication network according to claim 1, characterized in that, S4 include: The adjacent difference calculation is performed on the furnace pressure value sequence formed within the preset time period to obtain the difference sequence; Extract the sign of each element in the difference sequence, assign the positive difference to the first sign, the negative difference to the second sign, and the zero difference to the third sign, and arrange them in chronological order to form a sign sequence; The run length distribution is obtained by calculating the run lengths of each run consisting of consecutive identical symbols in a statistical symbol sequence. Tail feature parameters that characterize the tail shape of the run length distribution are extracted. These tail feature parameters include the maximum run length and the Pareto distribution shape parameter.

7. The automated troubleshooting and diagnosis method for vacuum furnaces based on a communication network according to claim 6, characterized in that, Extract the symbol of each element in the difference sequence, including: marking the positive element as the first symbol, the negative element as the second symbol, and the zero element as the third symbol, and arranging the marked symbols in the original time order of the difference sequence to form a symbol sequence.

8. The automated troubleshooting and diagnostic method for vacuum furnaces based on a communication network according to claim 6, characterized in that, Extracting tail-like feature parameters characterizing the tail morphology of the run-length distribution, including: Arrange the run length values ​​in the run length distribution in ascending order to obtain an ordered sequence of run lengths; Take the cumulative distribution function for the ordered sequence of run lengths and plot the cumulative distribution curve of the run length values ​​in a double logarithmic coordinate system; Take the data points in the tail region of the cumulative distribution curve. The tail region is the interval where the run length value is greater than the median run length value. P(L≥l) = (lmin / l)α α where l is the run length value, lmin is the minimum run length value in the tail region, a is the Pareto distribution shape parameter, and P(L≥l) is the cumulative probability of the run length value being greater than or equal to l. The shape parameter of the Pareto distribution and the maximum run length value in the ordered run length sequence are extracted and used together as the tail feature parameter.

9. The automated troubleshooting and diagnosis method for vacuum furnaces based on a communication network according to claim 1, characterized in that, S5 include: The recursion rate and determination rate in the recursive feature parameters are compared with the pre-stored coupling reference values, respectively. The maximum run length and Pareto distribution shape parameter in the tail feature parameters are compared with the pre-stored run reference values, respectively. When the maximum run length is less than the lower limit of the run length in the run reference value and the shape parameter of the Pareto distribution is greater than the upper limit of the shape parameter in the run reference value, and the recursion rate and the determination rate do not exceed the coupling reference value, it is determined that there is intermittent leakage or sealing defect in the furnace. When both the recursion rate and the determination rate exceed the coupling reference value, it is determined that there is a continuous leak in the furnace. When neither the recursion rate nor the determination rate exceeds the coupling reference value, and the maximum run length is greater than or equal to the lower limit of the run length and the Pareto distribution shape parameter is less than or equal to the upper limit of the shape parameter, the furnace is considered to be sealed normally.

10. The automated troubleshooting and diagnosis method for vacuum furnaces based on a communication network according to claim 1, characterized in that, S6 include: The results of the furnace leakage status determination are stored as a reference baseline; The valve connecting the furnace and the first bypass pipeline to be tested is opened by controlling the communication network. The overall execution after the furnace is connected to the first bypass pipeline to be tested is based on S2 to S5 to obtain the first leakage state result. The first leakage state result is compared with the reference standard. When the first leakage state result deteriorates relative to the reference standard, it is determined that the first bypass pipeline under test has a leak. Close the valve connecting the furnace to the first bypass pipeline to be tested, and open the valve connecting the furnace to the second bypass pipeline to be tested through the communication network. Repeat S2 to S5 for the entire furnace after it is connected to the second bypass pipeline to be tested to obtain the second leakage state result. The second leakage status result is compared with the reference standard. When the second leakage status result deteriorates relative to the reference standard, it is determined that there is a leak in the second bypass pipeline to be tested. Repeat the above process for the remaining bypass pipes to be tested until the leakage status of all bypass pipes to be tested is determined.

Images