An air conditioning system safety valve based on visual assistance

Abstract

The application discloses a kind of based on visual aid air conditioning system safety valve, belong to visual aid technical field, solve these problems existence certain concealment, for visual identification technology, it is difficult to judge dynamic pressure fluctuation condition by static image, and whether the rigidity of safety valve internal spring is appropriate Problem. Including safety valve body, three-axis vibration sensor being set on the valve body in safety valve body and collecting and outputting vibration signal, wideband pressure sensor being set in safety valve body entrance and collecting dynamic pressure change in safety valve body. The application detects the vibration and pressure condition of safety valve using three-axis vibration sensor and wideband pressure sensor, completes primary detection;Secondary detection is carried out by cooperating photoelectric pulse statistics and visual identification, and the final judgment is output combined with two-stage detection results, to ensure accurate identification of safety valve frequency jump or flutter, to avoid the influence of single visual identification technology on accuracy.

Description

Technical Field

[0001] This invention relates to the field of visual assistance technology, and in particular to a safety valve for an air conditioning system based on visual assistance. Background Technology

[0002] Air conditioning system safety valves are crucial components ensuring the safe operation of air conditioning equipment. Installed in the refrigeration system or on the compressor's exhaust pipe, their primary function is to limit system pressure, preventing damage from excessively high or low pressure. A safety valve consists of a valve body, valve core, spring, and adjusting nut. Normally closed, it automatically opens when the system pressure exceeds a set safety value, releasing excess pressure and restoring normal system pressure, thus protecting critical components such as the compressor and condenser from damage. It also prevents refrigerant leaks. In some air conditioning systems, safety valves are also equipped with pressure gauges for real-time system pressure monitoring by technicians. Their design is typically spring-loaded or lever-type, with the opening pressure set by adjusting the spring force or lever weight. The performance of the air conditioning system safety valve directly affects the safety and reliability of the system. Regular inspection and calibration of safety valves are essential for air conditioning maintenance, ensuring they function correctly in critical situations and guaranteeing stable system operation and user safety.

[0003] In the daily operation of air conditioning systems, frequent tripping or fluttering of safety valves is a common occurrence. One possible cause is excessive spring stiffness or improper adjustment ring positioning. When the spring stiffness is too high, after the valve disc opens, the large spring force causes it to quickly reseat. However, during this reseatment process, the collision between the valve disc and seat, as well as the impact of the fluid, can cause the valve disc to open again, repeating this cycle and leading to frequent tripping or fluttering. Improper adjustment ring positioning alters the safety valve's reseat pressure and opening / closing characteristics, preventing the valve disc from reseating stably after opening, thus causing fluttering. Another common cause is excessive back pressure fluctuation due to excessive resistance in the discharge pipe. When back pressure fluctuates, the force on the valve disc changes accordingly. If these fluctuations are frequent and significant, they can cause frequent tripping or fluttering of the valve disc.

[0004] However, these problems are somewhat hidden, making it difficult for visual recognition technology to judge dynamic pressure fluctuations and whether the stiffness of the spring inside the safety valve is appropriate from static images. Furthermore, in actual operating scenarios, complex pipeline layouts and obstructions further increase the difficulty of visual recognition, making it hard to identify and judge these problems in a timely and accurate manner.

[0005] Therefore, a vision-assisted safety valve for air conditioning systems is proposed to solve or alleviate the above problems. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by proposing a vision-assisted safety valve for air conditioning systems.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A vision-assisted safety valve for an air conditioning system includes a safety valve body, a triaxial vibration sensor mounted on the valve body of the safety valve body and for acquiring and outputting vibration signals, a broadband pressure sensor mounted inside the inlet of the safety valve body and for acquiring dynamic pressure changes within the safety valve body, a screen acquisition unit mounted on one side of the safety valve body and for covering the actuator within the safety valve body, and a photoelectric encoder that is drivenly connected to the actuator within the safety valve body.

[0009] It also includes a signal preprocessing module, a vibration spectrum separation module, a phase coherence detection module, a dynamic fault classification module, a visual motion verification module, and a communication and display module;

[0010] The input terminal of the signal preprocessing module is connected to the output terminal of the broadband pressure sensor, and the signal preprocessing module separates the dynamic pressure change into low-frequency, medium-frequency, and high-frequency channel signals and transmits them to the dynamic fault classification module and the phase coherence detection module.

[0011] The input of the vibration spectrum separation module is connected to the output of the triaxial vibration sensor, and the triaxial vibration sensor extracts the frequency band energy characteristics of the vibration signal to the dynamic fault classification module.

[0012] The output of the phase coherence detection module is connected to the input of the dynamic fault classification module. The phase coherence detection module receives the intermediate frequency channel signal and frequency band energy characteristics to calculate the phase difference between pressure fluctuation and vibration and sends it to the dynamic fault classification module.

[0013] The output of the dynamic fault classification module is connected to the input of the communication and display module, and the dynamic fault classification module outputs fault codes through logical decision-making.

[0014] The input terminal of the visual motion verification module is connected to the output terminal of the photoelectric encoder and the image acquisition unit, respectively. The output terminal of the visual motion verification module is also connected to the input terminal of the communication and display module. The visual motion verification module verifies the valve action frequency through visual data.

[0015] The communication and display module is used to receive fault codes and verification results and output graded alarms.

[0016] Preferably, the signal preprocessing module separates dynamic pressure changes into low-frequency, medium-frequency, and high-frequency channel signals and transmits them to the dynamic fault classification module and the phase coherence detection module, specifically including the following steps:

[0017] The pressure signal with dynamic pressure changes is separated by a third-order Sallen-Key filter;

[0018] The low-frequency channel signal uses a first-order low-pass filter with a frequency range of 0.1 to 2 Hz and a time constant of 0.8 seconds to detect slow changes in pipe resistance.

[0019] The intermediate frequency channel signal uses a second-order bandpass filter with a frequency range of 5 to 20 Hz, a center frequency of 10 Hz, and a quality factor of 0.707, which is used to extract the periodic fluctuations of the adjustment loop.

[0020] The high-frequency channel signal uses a second-order high-pass filter with a frequency range of 50 to 200 Hz, a center frequency of 50 Hz, and a quality factor of 0.5, to capture transient impacts on spring stiffness.

[0021] Preferably, the triaxial vibration sensor extracts the frequency band energy characteristics of the vibration signal, specifically including the following steps:

[0022] The vibration signal is separated by a programmable filter to obtain low-frequency energy characteristics and high-frequency energy characteristics. The frequency range of the low-frequency energy characteristics is 5 to 50 Hz, which corresponds to flutter detection. The frequency range of the high-frequency energy characteristics is 50 to 200 Hz, which corresponds to frequency hopping detection.

[0023] Preferably, the dynamic fault classification module outputs fault codes through logical decision-making, specifically including the following steps:

[0024] Based on the baseline tracking circuit in the dynamic fault classification module, the pressure baseline is calculated in real time and a dynamic threshold is generated.

[0025] In the dynamic fault classification module, pressure channel energy, vibration frequency band energy, and phase difference features are extracted and weighted fusion is performed.

[0026] The fault type is determined by the fuzzy decision circuit in the dynamic fault classification module, combined with threshold conditions and membership functions.

[0027] Preferably, the baseline tracking circuit in the dynamic fault classification module calculates the pressure baseline and generates a dynamic threshold in real time, specifically including the following steps:

[0028] The baseline tracking circuit in the dynamic fault classification module uses a first-order inertial filter. The current baseline voltage value is 95% of the previous baseline voltage value plus 5% of the current input voltage value.

[0029] In the dynamic fault classification module, the adaptive threshold circuit is implemented through a comparator. The dynamic threshold is the sum of the baseline voltage value and three times the standard deviation of the static noise, where the standard deviation of the static noise is determined by static calibration.

[0030] The parameters of the dynamic threshold are dynamically adjusted through the serial clock pin and data input pin of the digital potentiometer in the dynamic fault classification module.

[0031] Preferably, the step of extracting pressure channel energy, vibration frequency band energy, and phase difference features in the dynamic fault classification module, and then performing weighted fusion, specifically includes the following steps:

[0032] The input low-frequency channel signal, mid-frequency channel signal, high-frequency channel signal, low-frequency energy characteristics, and high-frequency energy characteristics are normalized.

[0033] After normalization, each signal is multiplied by a preset weighting coefficient and then summed to obtain a comprehensive fault score, where the weight of the low-frequency channel signal is 0.3, the weight of the mid-frequency channel signal is 0.25, the weight of the high-frequency channel signal is 0.25, the weight of the low-frequency energy characteristic is 0.1, and the weight of the high-frequency energy characteristic is 0.1.

[0034] Preferably, the step of determining the fault type by combining a threshold condition and a membership function through the fuzzy decision circuit in the dynamic fault classification module specifically includes the following steps:

[0035] The criteria for judging excessive spring stiffness are: the comprehensive fault score exceeds 1.5, the high-frequency channel signal exceeds three times its static standard deviation, and the phase difference between pressure and vibration is greater than 60 degrees.

[0036] The criteria for determining improper adjustment ring position are: the intermediate frequency channel signal exceeds twice its static standard deviation, and the ratio of the peak-to-peak value to the mean value of pressure fluctuation is greater than 0.4.

[0037] The criteria for determining excessive pipeline resistance are that the low-frequency channel signal exceeds 2.5 times its static standard deviation and the baseline voltage change rate exceeds 0.05 volts per second.

[0038] Preferably, the visual motion verification module verifies the valve's action frequency using visual data, specifically including the following steps:

[0039] The photoelectric encoder collects the rising edge of the photoelectric pulse to capture transient abnormalities, and the valve's operating frequency is the number of rising edges of the photoelectric pulse.

[0040] The camera captures images and calculates the displacement trend using the frame difference method. The displacement change is the average difference between the pixel grayscale values ​​of two adjacent frames. The threshold is set to 20 grayscale levels per frame. If the displacement trend exceeds 20 grayscale levels per frame, it is determined to be a valid action.

[0041] The action frequency is calculated based on the number of valid actions / total number of frames * time interval between two adjacent frames. If the valve action frequency - action frequency < 0.2 * valve action frequency, the output result is "verification passed".

[0042] Preferably, the communication and display module is used to receive fault codes and verification results and output graded alarms, specifically including the following steps:

[0043] If the fault type output by the dynamic fault classification module and the output result of the visual motion verification module are both received and the verification is passed, then information is transmitted externally and an alarm is triggered.

[0044] The present invention has the following beneficial effects:

[0045] This invention uses a triaxial vibration sensor and a wideband pressure sensor to detect the vibration and pressure of the safety valve body, thereby completing the first-level detection of frequency jumping or flutter in the safety valve body. In addition, it uses photoelectric pulse statistics to detect frequency jumping or flutter and visual recognition for second-level detection. The results of the first-level and second-level detections are combined to output the final result, ensuring that the safety valve body can be accurately identified when it experiences frequency jumping or flutter, and avoiding the impact of relying solely on visual recognition technology on the accuracy of the final problem identification. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram of the structure of the present invention.

[0048] In the diagram: 1. Safety valve body; 2. Wideband pressure sensor; 3. Triaxial vibration sensor; 4. Signal preprocessing module; 5. Vibration spectrum separation module; 6. Phase coherence detection module; 7. Dynamic fault classification module; 8. Visual motion verification module; 9. Communication and display module. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0050] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0051] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0052] In the description of this invention, it should be understood that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used to facilitate the description of this invention and to simplify the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0053] Furthermore, the terms "first," "second," and "third" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0054] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0055] A vision-assisted safety valve for an air conditioning system, such as Figure 1 As shown, it includes a safety valve body 1, a triaxial vibration sensor 3 installed on the valve body of the safety valve body 1 and collecting and outputting vibration signals, a broadband pressure sensor 2 installed in the inlet of the safety valve body 1 and collecting dynamic pressure changes in the safety valve body 1, a screen acquisition unit installed on one side of the safety valve body 1 and covering the field of view of the actuator in the safety valve body 1, and a photoelectric encoder that is connected to the actuator in the safety valve body 1. The screen acquisition unit is a camera.

[0056] It also includes a signal preprocessing module 4, a vibration spectrum separation module 5, a phase coherence detection module 6, a dynamic fault classification module 7, a visual motion verification module 8, and a communication and display module 9.

[0057] The input of the signal preprocessing module 4 is connected to the output of the broadband pressure sensor 2. The input of the vibration spectrum separation module 5 is connected to the output of the triaxial vibration sensor 3. The output of the phase coherence detection module 6 is connected to the input of the dynamic fault classification module 7. The output of the dynamic fault classification module 7 is connected to the input of the communication and display module 9. The input of the visual motion verification module 8 is connected to the output of the photoelectric encoder and the image acquisition unit, respectively. The output of the visual motion verification module 8 is also connected to the input of the communication and display module 9.

[0058] The signal preprocessing module 4 includes a DAC121S101 digital-to-analog converter, an INA333 instrumentation amplifier, a first resistor, an LTC1562 anti-aliasing filter, an OP07 operational amplifier, an LT1016 comparator, an LT1568 bandpass filter, a first AD633 multiplier, an AD827 differential amplifier, and an AD734 multiplier.

[0059] The wideband pressure sensor 2, model MS5803-14BA, has its serial clock and serial data pins connected to the serial clock and serial data inputs of the DAC121S101 digital-to-analog converter, respectively. The analog output pin of the DAC121S101 is connected to the non-inverting input of the INA333 instrumentation amplifier, which is grounded. A first resistor is connected in series between the two gain setting pins of the INA333 instrumentation amplifier. The output of the INA333 instrumentation amplifier is connected to the input of the LTC1562 anti-aliasing filter. The output of the LTC1562 anti-aliasing filter is divided into three paths:

[0060] Low-frequency channel: input via OP07 operational amplifier, the output of OP07 operational amplifier is connected to the non-inverting input of LT1016 comparator; Intermediate-frequency channel: input via the positive input of LT1568 bandpass filter, the output of LT1568 bandpass filter is connected to the X input of the first AD633 multiplier; High-frequency channel: input via the non-inverting input of AD827 differential amplifier, the output of AD827 differential amplifier is connected to the X input of AD734 multiplier.

[0061] The triaxial vibration sensor 3 is model ADXL345. The vibration spectrum separation module 5 includes an OPA2277 charge amplifier, an LTC1068 programmable filter, an AD736 RMS converter, a second resistor, a third resistor, a fourth resistor, a fifth resistor, a sixth resistor, and a seventh resistor.

[0062] The X-axis output of the triaxial vibration sensor 3 is connected to the non-inverting input of the OPA2277 charge amplifier. The output of the OPA2277 charge amplifier is connected to the input of the LTC1068 programmable filter. The six outputs of the LTC1068 programmable filter are connected to the six inputs of the AD736 RMS converter. The six outputs of the AD736 RMS converter are connected to the input of the dynamic fault classification module 7 after being connected in series with the second, third, fourth, fifth, sixth, and seventh resistors.

[0063] The phase coherence detection module 6 includes a CD4046 phase-locked loop, a second AD633 multiplier, and an OPA2188 integrator;

[0064] The input terminal of the CD4046 phase-locked loop is connected to the intermediate frequency channel output terminal of the signal preprocessing module 4. The output terminal of the CD4046 phase-locked loop is connected to the Y input terminal of the second AD633 multiplier. The X input terminal of the second AD633 multiplier is connected to the third channel output terminal of the vibration spectrum separation module 5. The Z input terminal of the second AD633 multiplier is connected to the input terminal of the OPA2188 integrator. The output terminal of the OPA2188 integrator outputs the phase difference voltage to the input terminal of the dynamic fault classification module 7.

[0065] The dynamic fault classification module 7 includes an AD734 eight-channel analog multiplier, a DS1804 digital potentiometer, an OPA211 limiting amplifier, and a three-channel LM311 comparator array.

[0066] The X input of the AD734 eight-channel analog multiplier is connected to the outputs of the low-frequency channel, the intermediate-frequency channel, the high-frequency channel, and the six outputs of the AD736 RMS converter. The Y input of the AD734 eight-channel analog multiplier is weighted through the serial clock and data inputs of the DS1804 digital potentiometer. The output of the AD734 eight-channel analog multiplier is connected to the non-inverting input of the OPA211 limiting amplifier. The output of the OPA211 limiting amplifier is connected to the inverting input of the three-channel LM311 comparator array. The non-inverting input of the three-channel LM311 comparator array is powered through the eighth resistor. The output of the three-channel LM311 comparator array is connected to the fault coding terminal of the communication and display module 9.

[0067] The visual motion verification module 8 includes an STM32F407 microcontroller and a TXB0108 level conversion chip;

[0068] The STM32F407 microcontroller is connected to the photoelectric encoder and the camera respectively. The clock pin and data output pin of the STM32F407 microcontroller are connected to the clock pin and data input pin of the communication and display module 9 after passing through the TXB0108 level conversion chip.

[0069] The communication and display module 9 includes a 74HC148 priority encoder and an RS485 driver chip. The enable terminal of the 74HC148 priority encoder is connected to the output terminal of the LM311 comparator in the dynamic fault classification module 7. The output terminal of the 74HC148 priority encoder is connected to the data input terminal of the RS485 driver chip. The differential output terminal of the RS485 driver chip is connected to an external RS485 bus to transmit information to the outside world.

[0070] Preferably, the signal preprocessing module 4 separates the dynamic pressure changes into low-frequency, medium-frequency, and high-frequency channel signals and transmits them to the dynamic fault classification module 7 and the phase coherence detection module 6, specifically including the following steps:

[0071] The pressure signal with dynamic pressure changes is separated by a third-order Sallen-Key filter;

[0072] The low-frequency channel signal uses a first-order low-pass filter with a frequency range of 0.1 to 2 Hz and a time constant of 0.8 seconds to detect slow changes in pipe resistance.

[0073] The intermediate frequency channel signal uses a second-order bandpass filter with a frequency range of 5 to 20 Hz, a center frequency of 10 Hz, and a quality factor of 0.707, which is used to extract the periodic fluctuations of the adjustment loop.

[0074] The high-frequency channel signal uses a second-order high-pass filter with a frequency range of 50 to 200 Hz, a center frequency of 50 Hz, and a quality factor of 0.5, to capture transient impacts on spring stiffness.

[0075] Preferably, the triaxial vibration sensor 3 extracts the frequency band energy characteristics of the vibration signal, specifically including the following steps:

[0076] The vibration signal is separated by a programmable filter to obtain low-frequency energy characteristics and high-frequency energy characteristics. The frequency range of the low-frequency energy characteristics is 5 to 50 Hz, which corresponds to flutter detection. The frequency range of the high-frequency energy characteristics is 50 to 200 Hz, which corresponds to frequency hopping detection.

[0077] Preferably, the dynamic fault classification module 7 outputs a fault code through logical decision-making, specifically including the following steps:

[0078] Based on the baseline tracking circuit in the dynamic fault classification module 7, the pressure baseline is calculated in real time and a dynamic threshold is generated.

[0079] In the dynamic fault classification module 7, the energy of the pressure channel, the energy of the vibration frequency band, and the phase difference features are extracted and weighted and fused.

[0080] The fault type is determined by the fuzzy decision circuit in the dynamic fault classification module 7, combined with threshold conditions and membership functions.

[0081] Preferably, based on the baseline tracking circuit in the dynamic fault classification module 7, the pressure baseline is calculated in real time and a dynamic threshold is generated, specifically including the following steps:

[0082] The baseline tracking circuit in the dynamic fault classification module 7 uses a first-order inertial filter. The current baseline voltage value is 95% of the previous baseline voltage value plus 5% of the current input voltage value.

[0083] In the dynamic fault classification module 7, the adaptive threshold circuit is implemented by a comparator. The dynamic threshold is the sum of the baseline voltage value and three times the standard deviation of static noise, where the standard deviation of static noise is determined by static calibration.

[0084] The parameters of the dynamic threshold are dynamically adjusted through the serial clock pin and data input pin of the digital potentiometer in the dynamic fault classification module 7.

[0085] Preferably, in the dynamic fault classification module 7, the pressure channel energy, vibration frequency band energy, and phase difference features are extracted and weighted fusion is performed, specifically including the following steps:

[0086] The input low-frequency channel signal, mid-frequency channel signal, high-frequency channel signal, low-frequency energy characteristics, and high-frequency energy characteristics are normalized.

[0087] After normalization, each signal is multiplied by a preset weighting coefficient and then summed to obtain a comprehensive fault score, where the weight of the low-frequency channel signal is 0.3, the weight of the mid-frequency channel signal is 0.25, the weight of the high-frequency channel signal is 0.25, the weight of the low-frequency energy characteristic is 0.1, and the weight of the high-frequency energy characteristic is 0.1.

[0088] Preferably, the fault type is determined by the fuzzy decision circuit in the dynamic fault classification module 7, combining threshold conditions and membership functions, specifically including the following steps:

[0089] The criteria for judging excessive spring stiffness are: the comprehensive fault score exceeds 1.5, the high-frequency channel signal exceeds three times its static standard deviation, and the phase difference between pressure and vibration is greater than 60 degrees.

[0090] The criteria for determining improper adjustment ring position are: the comprehensive fault score exceeds 1.8, the intermediate frequency channel signal exceeds twice its static standard deviation, and the ratio of the peak-to-peak value of pressure fluctuation to the mean value is greater than 0.4.

[0091] The criteria for determining excessive pipeline resistance are: a comprehensive fault score exceeding 2.2, a low-frequency channel signal exceeding 2.5 times its static standard deviation, and a baseline voltage change rate exceeding 0.05 volts per second.

[0092] Preferably, the visual motion verification module 8 verifies the valve's action frequency through visual data, specifically including the following steps:

[0093] The photoelectric encoder collects the rising edge of the photoelectric pulse to capture transient abnormalities, and the valve's operating frequency is the number of rising edges of the photoelectric pulse.

[0094] The camera captures images and calculates the displacement trend using the frame difference method. The displacement change is the average difference between the pixel grayscale values ​​of two adjacent frames. The threshold is set to 20 grayscale levels per frame. If the displacement trend exceeds 20 grayscale levels per frame, it is determined to be a valid action.

[0095] The action frequency is calculated based on the number of valid actions / total number of frames * time interval between two adjacent frames. If the valve action frequency - action frequency < 0.2 * valve action frequency, the output result is "verification passed".

[0096] Preferably, the communication and display module 9 is used to receive fault codes and verification results and output graded alarms, specifically including the following steps:

[0097] If the fault type output by the dynamic fault classification module 7 and the output result of the visual motion verification module 8 are both received and the verification is passed, then information is transmitted externally and an alarm is triggered.

[0098] When the air conditioning system is started, the wideband pressure sensor 2 collects the pressure data at the inlet side of the safety valve in real time. The digital signal is transmitted to the DAC121S101 digital-to-analog converter, converted into a 0-5V analog signal, and then output to the INA333 instrumentation amplifier. After gain amplification, the signal is output to the LTC1562 anti-aliasing filter. This filter decomposes the pressure signal into three frequency bands through a third-order Sallen-Key structure: the low-frequency channel signal is processed by the input and output of the OP07 operational amplifier, and the pressure change caused by the resistance at the inlet side of the safety valve is detected through integration. Its output is connected to the LT1016 comparator and compared with the dynamic threshold.

[0099] The intermediate frequency channel signal is fed into the X input terminal of the first AD633 multiplier after the periodic fluctuation characteristics of the adjustment loop are extracted from the input and output of the LT1568 bandpass filter.

[0100] The high-frequency channel signal is input and output through the AD827 differential amplifier to capture the high-frequency impact signal with excessive spring stiffness and input to the X input terminal of the AD734 multiplier.

[0101] Meanwhile, the X-axis output of the triaxial vibration sensor 3 is connected to the OPA2277 charge amplifier. After amplification, the vibration signal is output to the LTC1068 programmable filter. The filter separates the vibration signal into low-frequency and high-frequency bands. The low-frequency band is output to the AD736 RMS converter to detect flutter energy, and the high-frequency band is output to the AD736 RMS converter to detect frequency jump energy. The six RMS signals are input to the dynamic fault classification module 7 via resistor networks.

[0102] During the dynamic threshold generation stage, the dynamic fault classification module 7 calculates the pressure baseline voltage in real time using a first-order inertial filtering algorithm. The baseline value is the sum of 95% of the previous baseline voltage and 5% of the current input voltage, input to the OPA2188 integrator, and output to the LM311 comparator. The generated dynamic threshold is the baseline value plus three times the static noise standard deviation, and the threshold signal is input to a three-channel LM311 comparator array.

[0103] In the fault feature extraction and fusion stage, the AD734 analog multiplier receives low-frequency, medium-frequency, high-frequency pressure signals, low-frequency vibration signals, and high-frequency vibration signals, respectively. The weights of each channel are set by the DS1804 digital potentiometer, with weights of 0.3, 0.25, 0.25, 0.1, and 0.1, respectively. The multiplication result is output to the OPA211 limiting amplifier, and after amplification, the comprehensive fault score is output to the three-channel LM311 comparator array, which compares it with the preset threshold to generate a 3-bit fault code.

[0104] In the phase coherence detection module 6, the CD4046 phase-locked loop receives the pressure intermediate frequency signal, and the output VCO signal is input to the Y input terminal of the second AD633 multiplier. After being multiplied with the vibration third channel signal, the OPA2188 integrator receives the product signal and outputs the phase difference voltage to the dynamic fault classification module 7, which is used to determine the pressure-vibration coherence.

[0105] In the visual motion verification module 8, the output of the photoelectric encoder is connected to the STM32F407 microcontroller. The valve action frequency is calculated by counting the rising edges. At the same time, the camera acquires images of the valve stem at a rate of 10 frames per second. The average grayscale difference between adjacent frames is calculated by the frame difference method. If it exceeds 20 grayscale levels, it is determined to be a valid action. The verification result is transmitted to the communication and display module 9 through the SPI interface via the TXB0108 level conversion chip.

[0106] In the communication and display module 9, the 74HC148 priority encoder is enabled to receive fault codes and outputs encoded signals to the RS485 driver chip, driving the RS485 bus to send differential signals.

[0107] Throughout the process, multi-sensor data is strictly synchronized through timestamps. The frequency band separation of pressure signals, the spectrum analysis of vibration energy, the coherence calculation of phase difference, and the visual motion verification form a quadruple redundancy. Dynamic thresholds avoid environmental interference, and fuzzy decision logic ensures a low false alarm rate. From pressure impact capture and mechanical vibration analysis to motion trajectory verification, it fully covers the detection needs of three core faults: abnormal spring stiffness, adjustment ring offset, and excessive pipeline resistance, avoiding situations where visual recognition technology alone cannot accurately determine the problem.

[0108] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A safety valve for an air conditioning system based on vision assistance, characterized in that, Includes a safety valve body (1), a triaxial vibration sensor (3) installed on the valve body in the safety valve body (1) and collecting and outputting vibration signals, a broadband pressure sensor (2) installed in the inlet of the safety valve body (1) and collecting dynamic pressure changes in the safety valve body (1), a screen acquisition unit installed on one side of the safety valve body (1) and covering the field of view of the actuator in the safety valve body (1), and a photoelectric encoder that is connected to the actuator in the safety valve body (1) for transmission. It also includes a signal preprocessing module (4), a vibration spectrum separation module (5), a phase coherence detection module (6), a dynamic fault classification module (7), a visual motion verification module (8), and a communication and display module (9); The input end of the signal preprocessing module (4) is connected to the output end of the wideband pressure sensor (2), and the signal preprocessing module (4) separates the dynamic pressure change into low frequency, medium frequency and high frequency channel signals and transmits them to the dynamic fault classification module (7) and the phase coherence detection module (6). The input end of the vibration spectrum separation module (5) is connected to the output end of the triaxial vibration sensor (3), and the triaxial vibration sensor (3) extracts the frequency band energy characteristics of the vibration signal to the dynamic fault classification module (7). The output of the phase coherence detection module (6) is connected to the input of the dynamic fault classification module (7). The phase coherence detection module (6) receives the intermediate frequency channel signal and frequency band energy characteristics to calculate the phase difference between pressure fluctuation and vibration and transmits it to the dynamic fault classification module (7). The output of the dynamic fault classification module (7) is connected to the input of the communication and display module (9), and the dynamic fault classification module (7) outputs fault codes through logical decision-making. The input end of the visual motion verification module (8) is connected to the output end of the photoelectric encoder and the image acquisition unit respectively. The output end of the visual motion verification module (8) is also connected to the input end of the communication and display module (9). The visual motion verification module (8) is used to verify the valve action frequency through visual data. The communication and display module (9) is used to receive fault codes and verification results and output graded alarms.

2. The safety valve for an air conditioning system based on vision assistance according to claim 1, characterized in that, The signal preprocessing module (4) separates the dynamic pressure change into low-frequency, medium-frequency, and high-frequency channel signals and transmits them to the dynamic fault classification module (7) and the phase coherence detection module (6), specifically including the following steps: The pressure signal with dynamic pressure changes is separated by a third-order Sallen-Key filter; The low-frequency channel signal uses a first-order low-pass filter with a frequency range of 0.1 to 2 Hz and a time constant of 0.8 seconds to detect slow changes in pipe resistance. The intermediate frequency channel signal uses a second-order bandpass filter with a frequency range of 5 to 20 Hz, a center frequency of 10 Hz, and a quality factor of 0.707, which is used to extract the periodic fluctuations of the adjustment loop. The high-frequency channel signal uses a second-order high-pass filter with a frequency range of 50 to 200 Hz, a center frequency of 50 Hz, and a quality factor of 0.5, to capture transient impacts on spring stiffness.

3. The safety valve for an air conditioning system based on vision assistance according to claim 1, characterized in that, The triaxial vibration sensor (3) extracts the frequency band energy characteristics of the vibration signal, specifically including the following steps: The vibration signal is separated by a programmable filter to obtain low-frequency energy characteristics and high-frequency energy characteristics. The frequency range of the low-frequency energy characteristics is 5 to 50 Hz, which corresponds to flutter detection. The frequency range of the high-frequency energy characteristics is 50 to 200 Hz, which corresponds to frequency hopping detection.

4. A vision-assisted safety valve for an air conditioning system according to claim 1, characterized in that, The dynamic fault classification module (7) outputs fault codes through logical decision-making, specifically including the following steps: Based on the baseline tracking circuit in the dynamic fault classification module (7), the pressure baseline is calculated in real time and a dynamic threshold is generated. In the dynamic fault classification module (7), the energy of the pressure channel, the energy of the vibration frequency band and the phase difference features are extracted and weighted and fused. The fault type is determined by the fuzzy decision circuit in the dynamic fault classification module (7), combined with the threshold condition and membership function.

5. A vision-assisted safety valve for an air conditioning system according to claim 4, characterized in that, The baseline tracking circuit in the dynamic fault classification module (7) calculates the pressure baseline and generates a dynamic threshold in real time, specifically including the following steps: The baseline tracking circuit in the dynamic fault classification module (7) uses first-order inertial filtering. The current baseline voltage value is 95% of the previous baseline voltage value plus 5% of the current input voltage value. In the dynamic fault classification module (7), the adaptive threshold circuit is implemented by a comparator. The dynamic threshold is the sum of the baseline voltage value and three times the standard deviation of static noise, where the standard deviation of static noise is determined by static calibration. The parameters of the dynamic threshold are dynamically adjusted through the serial clock pin and data input pin of the digital potentiometer in the dynamic fault classification module (7).

6. A vision-assisted safety valve for an air conditioning system according to claim 4, characterized in that, In the dynamic fault classification module (7), the energy of the pressure channel, the energy of the vibration frequency band, and the phase difference features are extracted and weighted and fused. The specific steps include the following: The input low-frequency channel signal, mid-frequency channel signal, high-frequency channel signal, low-frequency energy characteristics, and high-frequency energy characteristics are normalized. After normalization, each signal is multiplied by a preset weighting coefficient and then summed to obtain a comprehensive fault score, where the weight of the low-frequency channel signal is 0.3, the weight of the mid-frequency channel signal is 0.25, the weight of the high-frequency channel signal is 0.25, the weight of the low-frequency energy characteristic is 0.1, and the weight of the high-frequency energy characteristic is 0.

1.

7. A vision-assisted safety valve for an air conditioning system according to claim 4, characterized in that, The process of determining the fault type by combining threshold conditions and membership functions through the fuzzy decision circuit in the dynamic fault classification module (7) includes the following steps: The criteria for judging excessive spring stiffness are: the comprehensive fault score exceeds 1.5, the high-frequency channel signal exceeds three times its static standard deviation, and the phase difference between pressure and vibration is greater than 60 degrees. The criteria for determining improper adjustment ring position are: the intermediate frequency channel signal exceeds twice its static standard deviation, and the ratio of the peak-to-peak value to the mean value of pressure fluctuation is greater than 0.

4. The criteria for determining excessive pipeline resistance are that the low-frequency channel signal exceeds 2.5 times its static standard deviation and the baseline voltage change rate exceeds 0.05 volts per second.

8. A vision-assisted safety valve for an air conditioning system according to claim 1, characterized in that, The visual motion verification module (8) is used to verify the valve action frequency through visual data, specifically including the following steps: The photoelectric encoder collects the rising edge of the photoelectric pulse to capture transient abnormalities, and the valve's operating frequency is the number of rising edges of the photoelectric pulse. The camera captures images and calculates the displacement trend using the frame difference method. The displacement change is the average difference between the pixel grayscale values ​​of two adjacent frames. The threshold is set to 20 grayscale levels per frame. If the displacement trend exceeds 20 grayscale levels per frame, it is determined to be a valid action. The action frequency is calculated based on the number of valid actions / total number of frames * time interval between two adjacent frames. If the valve action frequency - action frequency < 0.2 * valve action frequency, the output result is "verification passed".

9. A vision-assisted safety valve for an air conditioning system according to claim 1, characterized in that, The communication and display module (9) is used to receive fault codes and verification results and output graded alarms, specifically including the following steps: If the fault type output by the dynamic fault classification module (7) and the output result of the visual motion verification module (8) are verified as passed, then the information is transmitted to the outside and an alarm is triggered.

Images