Molecular sieve oxygen generator alarm detection system and method

Abstract

This invention relates to the technical field of alarm devices and status monitoring systems, and discloses an alarm detection system and method for a molecular sieve oxygen generator, comprising: a pressure signal conditioning unit that collects pressure data sequences during a closed pressurization period and calculates the first-order time-series difference value, and outputs pressure change parameters; a current signal conditioning unit that extracts the rate of change of the fundamental amplitude envelope of the current power frequency signal and outputs current change parameters; an alarm status determination unit that constructs two-dimensional data points based on the pressure change parameters and current change parameters, and determines that an abnormal state is established when the abnormal state exceeds the composite alarm boundary line; and an alarm signal output unit that changes the pin level state and outputs an alarm signal when the number of consecutive cycles of the abnormal state reaches a threshold. This invention, by coordinating the pressure and electromagnetic load characteristics of the pneumatic circuit, eliminates voltage fluctuations and start-up load interference, accurately identifies bed damage, and reduces the false alarm rate.

Description

Technical Field

[0001] This invention belongs to the technical field of alarm devices and status monitoring systems, and particularly relates to an alarm detection system and method for a molecular sieve oxygen generator. Background Technology

[0002] Current gas separation equipment operation status monitoring systems typically install a gas composition analyzer at the end of the oxygen production pipeline or a pressure control switch on the adsorption tower inlet pipeline. A comparison chip is used to retrieve discretely collected pressure amplitudes or oxygen purity fluctuations. By comparing the collected data with fixed safety boundary parameters, the drive module issues an abnormality warning when the value exceeds a preset range. Mobile pressure swing adsorption (PSA) systems face variable environmental loads. In the initial cold start-up phase, the viscosity of the compressor lubricating oil increases resistance. During the pressurization phase, brief pressure drop spikes often occur, and transient fluctuations in the external grid voltage further exacerbate the problem. The time-varying distortion of compressor exhaust efficiency often leads to false alarms in a healthy adsorption circuit due to slower pressure build-up. In the early stages of structural degradation of the molecular sieve adsorption bed, the waveform distortion caused by changes in gas flow resistance is very weak. Conventional static threshold comparison schemes lack the ability to dynamically analyze the periodic time sequence, resulting in a severe lag in response. The output is only triggered when the oxygen purity at the end decreases significantly. If the sensitivity is improved by narrowing the fixed boundary, the tolerance of the alarm system to start-up load and voltage fluctuations is reduced, thus causing the entire circuit to fall into a dilemma of increasing false alarm rate and increased risk of missed alarms.

[0003] Hardware sampling and absolute thresholds have limitations, and control methods also have shortcomings. For example, Chinese invention patent application CN116026398A discloses a molecular sieve failure detection method, device, equipment, and storage medium. It determines failure by acquiring the inlet pressure and compressor temperature and mapping the relationship between the two based on the ideal gas law. This architecture relies on static thermodynamic equilibrium. In practical applications, oxygen generators are accompanied by dynamic load disturbances such as transient voltage changes in the power grid. The compressor temperature change has thermal inertia and temperature rise lag, which cannot map the sudden change in exhaust slope caused by the input voltage drop in real time. This results in a fundamental mismatch between the reference pressure derived from the state equation and the transient change in operating conditions. The control loop cannot offset the power supply noise and frequently causes false alarms, making it impossible to accurately locate the surface layer of the sieve structure damage.

[0004] Therefore, the technical problem to be solved by this invention is how to integrate the transient pressure response waveform of the pneumatic circuit with the current load characteristics of the power supply side, and construct a graded alarm judgment system and alarm signal output control logic for feedforward sensing of hidden damage to the adsorption bed under the condition of eliminating initial mechanical friction, transient fluctuation of grid voltage and interference of valve body natural wear and leakage. Summary of the Invention

[0005] The present invention aims to solve the problem that existing alarm devices have high false alarm rates and missed alarm rates under complex alternating operating conditions due to their single decision dimension and lack of adaptive analysis capability for dynamic time sequence characteristics.

[0006] In this technical solution, an alarm detection system for a molecular sieve oxygen generator includes:

[0007] The pressure signal conditioning unit is connected to the alarm status determination unit, the current signal conditioning unit is connected to the alarm status determination unit, and the alarm status determination unit is connected to the alarm signal output unit. The pressure signal conditioning unit collects the pressure data sequence of the closed pressurization period of the pneumatic circuit in the molecular sieve pressure swing adsorption cycle, calculates the first-order time difference value of the pressure data sequence, and outputs the pressure change parameter.

[0008] The current signal conditioning unit acquires the power frequency signal driving the compressor, extracts the rate of change of the fundamental amplitude envelope of the power frequency signal, and outputs the current change parameter.

[0009] The alarm status determination unit receives pressure change parameters and current change parameters, constructs two-dimensional data points characterizing the system's operating features, and determines that the molecular sieve abnormal state is established when the two-dimensional data points exceed the set composite alarm boundary line; the alarm signal output unit changes the output level state of the main control chip pin when the number of consecutive cycles of the abnormal state reaches the set counting threshold, generating a jump alarm signal.

[0010] Preferably, the system further includes a leakage deviation compensation unit; the leakage deviation compensation unit is connected to the alarm status determination unit, collects the lowest valley pressure value during the desorption period to determine the leakage value of the distribution valve, and converts the leakage value into a correction weight inside the alarm status determination unit, and shifts the composite alarm boundary line according to the change in the correction weight in each pressure swing adsorption cycle.

[0011] Preferably, the system further includes a frequency domain filtering module; the frequency domain filtering module is connected between the pressure signal conditioning unit and the alarm signal output unit, calculates the amplitude integral energy value of the pressure data sequence in a preset high frequency band, and adjusts the set counting threshold in the alarm signal output unit to the multi-sampling period accumulated value when the amplitude integral energy value is greater than the preset vibration threshold value.

[0012] Preferably, the alarm signal output unit includes a counter and a status register connected to the counter; when the counter receives a composite alarm boundary line exceeding the judgment signal within 2405 consecutive pressure swing adsorption cycles, the abnormal state is finally confirmed, and the stored value of the status register is changed, driving the output level state of the main control chip pin to switch to a level transition signal.

[0013] Preferably, the pressure signal conditioning unit calculates the pressure change parameters by including the following steps: acquiring the pressure transient waveforms of the pneumatic circuit during the pressurization and stabilization phases, and discretizing the pressure transient waveforms into a pressure data sequence; calculating the discrete point difference of the pressure data sequence, and extracting the value representing the maximum slope of the pressurization process as the pressure change parameter.

[0014] Preferably, the current signal conditioning unit extracts the current change parameter by including the following steps: acquiring the fundamental component signal in the current power frequency signal; extracting the amplitude envelope of the fundamental component signal, and calculating the rate of change of the amplitude envelope with time as the current change parameter.

[0015] Preferably, when the alarm signal output unit generates a jump alarm signal, it includes the following steps: at the end of each pressure swing adsorption cycle, the judgment result output by the alarm state judgment unit is accumulated to obtain the number of consecutive valid cycles; when the number of consecutive valid cycles reaches 2405 consecutive sampling cycles and all abnormal states are valid, the output level state of the main control chip pin is changed.

[0016] Preferably, the system also includes an audible and visual alarm response module; the audible and visual alarm response module is connected to the alarm signal output unit and includes a light-emitting diode and a speaker; the light-emitting diode receives a jump alarm signal in an abnormal state and flashes at a frequency of 3Hz; the speaker outputs an intermittent audio alarm sound.

[0017] Preferably, the system also includes a power supply monitoring unit; the power supply monitoring unit is connected to the alarm status determination unit and collects grid-side power supply input voltage data; when the power supply input voltage data is lower than the set power supply threshold and the pressure change parameter decreases monotonically, the alarm status determination unit keeps the composite alarm boundary line within the current set range.

[0018] A method for detecting an alarm in a molecular sieve oxygen generator, used in operating an alarm detection system for a molecular sieve oxygen generator, comprising:

[0019] Step S1: Collect the pressure data sequence of the closed pressurization period of the pneumatic loop in the molecular sieve pressure swing adsorption cycle through the pressure signal conditioning unit, calculate the first-order time difference value of the pressure data sequence, and output the pressure change parameter.

[0020] Step S2: Acquire the power frequency signal of the current driving the compressor through the current signal conditioning unit, extract the rate of change of the fundamental amplitude envelope of the power frequency signal, and output the current change parameter.

[0021] Step S3: The alarm status determination unit receives pressure change parameters and current change parameters, constructs two-dimensional data points characterizing the system operation characteristics, and determines that the molecular sieve abnormal state is established when the two-dimensional data points exceed the set composite alarm boundary line.

[0022] Step S4: When the number of consecutive cycles of an abnormal state is established reaches a set counting threshold, the alarm signal output unit changes the output level of the main control chip pin to generate a jump alarm signal.

[0023] Compared with existing technologies, the molecular sieve oxygen generator alarm detection system of the present invention has the following advantages:

[0024] 1. In the alarm detection of molecular sieve oxygen generators, by synergistically integrating the pressure timing difference value of the pressure swing adsorption pipeline during the pressurization transient stage with the rate of change of the fundamental envelope of the current on the compressor power supply side, a virtual phase space with gas flow characteristics and electromagnetic load characteristics as orthogonal axes is constructed. Utilizing the bidirectional deflection law in the spatial coordinates caused by the reduction in flow resistance due to molecular sieve pulverization and the reduction in compressor workload, the interference of slowed pressure build-up caused by the drop in input voltage on the external network side is eliminated. Under the technical requirements of ensuring oxygen safety, the system status judgment is freed from the limitations of traditional static pressure threshold over-limit comparison, providing a fault feature feedforward identification channel with causal logic chain support, accurately locating internal structural damage in the gas path.

[0025] 2. By setting a start-up duration counter inside the pressure signal conditioning unit, the cold-running mode logic is activated during the initial power-on period. This causes the internal alarm status judgment unit to adaptively expand the boundary tolerance radius of the fault convergence closed area set in the virtual phase space. This actively avoids transient non-technical pseudo-electromagnetic distortion and pipeline pressure oscillation caused by low temperature environment or excessively high compressor lubricating oil viscosity. It also eliminates sudden noise interference on the alarm judgment result during the start-up pressure establishment stage, and avoids misjudging this normal initial physical fluctuation as adsorption bed damage. As a result, the control status output of the entire alarm judgment system accurately converges to the actual physical structure damage of the bed.

[0026] 3. By continuously monitoring the lowest valley pressure during the desorption period through the added dynamic leakage resistance offset compensation unit, the trace leakage characteristics caused by long-term impact wear of the distribution valve are extracted and converted into the correction weight of the internal boundary translation operator of the alarm state judgment unit. The boundary of the fault convergence closed area in the virtual space is translated synchronously with the mechanical aging law of the equipment itself. Under the premise of ensuring that the natural wear of the valve body does not produce false false alarms, the judgment standard has the ability to reverse self-calibrate against the inherent leakage deviation of the gas circuit, thereby eliminating the nonlinear interactive interference caused by the local air tightness reduction of the distribution valve component on the judgment of the molecular sieve deterioration state. Attached Figure Description

[0027] Figure 1 This is a flowchart of an alarm detection method for a molecular sieve oxygen generator according to the present invention;

[0028] Figure 2 This is a structural diagram of an alarm detection system for a molecular sieve oxygen generator according to the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0030] An alarm detection system for a molecular sieve oxygen generator includes:

[0031] The pressure signal conditioning unit is connected to the alarm status determination unit, the current signal conditioning unit is connected to the alarm status determination unit, and the alarm status determination unit is connected to the alarm signal output unit. The pressure signal conditioning unit collects the pressure data sequence of the closed pressurization period of the pneumatic circuit in the molecular sieve pressure swing adsorption cycle, calculates the first-order time difference value of the pressure data sequence, and outputs the pressure change parameter.

[0032] The current signal conditioning unit acquires the power frequency signal driving the compressor, extracts the rate of change of the fundamental amplitude envelope of the power frequency signal, and outputs the current change parameter.

[0033] The alarm status determination unit receives pressure change parameters and current change parameters, constructs two-dimensional data points characterizing the system's operating features, and determines that the molecular sieve abnormal state is established when the two-dimensional data points exceed the set composite alarm boundary line; the alarm signal output unit changes the output level state of the main control chip pin when the number of consecutive cycles of the abnormal state reaches the set counting threshold, generating a jump alarm signal.

[0034] Preferably, the system further includes a leakage deviation compensation unit; the leakage deviation compensation unit is connected to the alarm status determination unit, collects the lowest valley pressure value during the desorption period to determine the leakage value of the distribution valve, and converts the leakage value into a correction weight inside the alarm status determination unit, and shifts the composite alarm boundary line according to the change in the correction weight in each pressure swing adsorption cycle.

[0035] Preferably, the system further includes a frequency domain filtering module; the frequency domain filtering module is connected between the pressure signal conditioning unit and the alarm signal output unit, calculates the amplitude integral energy value of the pressure data sequence in a preset high frequency band, and adjusts the set counting threshold in the alarm signal output unit to the multi-sampling period accumulated value when the amplitude integral energy value is greater than the preset vibration threshold value.

[0036] Preferably, the alarm signal output unit includes a counter and a status register connected to the counter; when the counter receives a composite alarm boundary line exceeding the judgment signal within 2405 consecutive pressure swing adsorption cycles, the abnormal state is finally confirmed, and the stored value of the status register is changed, driving the output level state of the main control chip pin to switch to a level transition signal.

[0037] Preferably, the pressure signal conditioning unit calculates the pressure change parameters by including the following steps: acquiring the pressure transient waveforms of the pneumatic circuit during the pressurization and stabilization phases, and discretizing the pressure transient waveforms into a pressure data sequence; calculating the discrete point difference of the pressure data sequence, and extracting the value representing the maximum slope of the pressurization process as the pressure change parameter.

[0038] Preferably, the current signal conditioning unit extracts the current change parameter by including the following steps: acquiring the fundamental component signal in the current power frequency signal; extracting the amplitude envelope of the fundamental component signal, and calculating the rate of change of the amplitude envelope with time as the current change parameter.

[0039] Preferably, when the alarm signal output unit generates a jump alarm signal, it includes the following steps: at the end of each pressure swing adsorption cycle, the judgment result output by the alarm state judgment unit is accumulated to obtain the number of consecutive valid cycles; when the number of consecutive valid cycles reaches 2405 consecutive sampling cycles and all abnormal states are valid, the output level state of the main control chip pin is changed.

[0040] Preferably, the system also includes an audible and visual alarm response module; the audible and visual alarm response module is connected to the alarm signal output unit and includes a light-emitting diode and a speaker; the light-emitting diode receives a jump alarm signal in an abnormal state and flashes at a frequency of 3Hz; the speaker outputs an intermittent audio alarm sound.

[0041] Preferably, the system also includes a power supply monitoring unit; the power supply monitoring unit is connected to the alarm status determination unit and collects grid-side power supply input voltage data; when the power supply input voltage data is lower than the set power supply threshold and the pressure change parameter decreases monotonically, the alarm status determination unit keeps the composite alarm boundary line within the current set range.

[0042] A method for detecting an alarm in a molecular sieve oxygen generator, comprising:

[0043] Step S1: Collect the pressure data sequence of the closed pressurization period of the pneumatic loop in the molecular sieve pressure swing adsorption cycle through the pressure signal conditioning unit, calculate the first-order time difference value of the pressure data sequence, and output the pressure change parameter.

[0044] Step S2: Acquire the power frequency signal of the current driving the compressor through the current signal conditioning unit, extract the rate of change of the fundamental amplitude envelope of the power frequency signal, and output the current change parameter.

[0045] Step S3: The alarm status determination unit receives pressure change parameters and current change parameters, constructs two-dimensional data points characterizing the system operation characteristics, and determines that the molecular sieve abnormal state is established when the two-dimensional data points exceed the set composite alarm boundary line.

[0046] Step S4: When the number of consecutive cycles of an abnormal state is established reaches a set counting threshold, the alarm signal output unit changes the output level of the main control chip pin to generate a jump alarm signal.

[0047] Example 1: The alarm detection system of this invention is integrated into a microcontroller chip with floating-point arithmetic capabilities. It is used to perform system-level status assessment for an alarm detection system of a molecular sieve oxygen generator. This alarm detection system technically belongs to the G08B class of alarm devices. During the continuous operation of the molecular sieve oxygen generator in the long-term life support environment, there are alternating pressure swing adsorption shocks and aeroacoustic noise in the pneumatic circuit. Traditional alarm devices based on terminal oxygen concentration or static pressure thresholds are easily affected by mechanical resistance caused by the viscosity of lubricating oil during the initial cold start of the compressor, or by transient fluctuations in the bias voltage of the external power grid, frequently generating invalid false alarms. This leads to negligence by oxygen users, resulting in secondary damage to components due to the failure to report system faults in a timely manner. The alarm detection system of this invention, without changing the basic sensing hardware structure of the molecular sieve oxygen generator, directly acts on the physical cycle characteristic variables. It uses a pressure signal conditioning unit to collect the pressure data sequence during the closed pressurization period of the pneumatic circuit within the molecular sieve pressure swing adsorption cycle, calculates the first-order temporal difference value of the pressure data sequence, and outputs the first characteristic parameter. The current signal conditioning unit simultaneously acquires the power frequency signal driving the compressor, extracts the rate of change of the fundamental amplitude envelope of the power frequency signal, and outputs the second characteristic parameter. Under fatigue conditions after 2400 hours of operation, the molecular sieve oxygen generator experiences pulverization due to the high-frequency mechanical impact of long-term reversing airflow on some sieve bodies. This leads to a decrease in gas flow resistance. This decrease in flow resistance is attributed to the channeling effect formed by the high-frequency impact of airflow after the molecular sieve bed pulverization. Specifically, the locally pulverized particles are lost with the airflow or accumulate at the bottom of the bed, resulting in non-uniform physical voids or airflow short-circuit channels between the originally tightly packed molecular sieve crystals. Compared to the limited micropore diffusion resistance under healthy conditions, the frictional resistance experienced by the gas flowing through these overall cracks formed by structural failure is reduced. Overall, this manifests as a decrease in the rate of pressure rise within the tower during the closed pressurization period. The abnormal acceleration, a characteristic of reduced resistance, contrasts with the slower pressure build-up caused by compressor mechanical wear, forming a physical causal difference. This difference constitutes the underlying criterion for accurately identifying bed damage in this scheme. During the 2401st pressurization cycle, the pressure signal conditioning unit captures data on the pressurized phase of the adsorption tower at a fixed sampling frequency of 50Hz. Within a discrete pressurization sampling window with a 20ms delay relative to the switching solenoid valve's trigger signal and a duration spanning 15ms, the discrete sequence of intake pressure is acquired and a first-order time-series difference calculation is performed. The specific calculation formula is as follows: ,in, The first characteristic parameter characterizes the transient pressure build-up slope under pressure. for The pressure signal of the pneumatic circuit at all times. To fix the discrete time step at 20 ms, the first characteristic parameter is calculated because the pore closure of the molecular sieve reduces the resistance to airflow. A monotonic step increase is generated, with the value continuously rising and stabilizing at 2.49. Simultaneously, the current signal conditioning unit uses a sampling shunt to acquire the power frequency signal from the compressor's power input at a fixed sampling rate of 1kHz. It extracts the fundamental component signal and its amplitude envelope within the same sampling window, calculates the rate of change of the amplitude envelope over time, and outputs the second characteristic parameter. Because the exhaust backflow resistance decreases, the output load of the compressor motor becomes lighter, and the current signal conditioning unit calculates the second characteristic parameter. The value dropped, monotonically sliding down to 0.035.

[0048] The alarm status determination unit receives the first characteristic parameter in real time. With the second characteristic parameter , construct With the horizontal axis as the base, The vertical axis represents a two-dimensional virtual control topology phase space, and the feature parameters extracted in real time are grouped together. Mapped to dynamic topology nodes within space, the system has a pre-defined composite alarm boundary line formed by expanding a baseline rectangular safety area obtained under standard healthy operating conditions, stored in non-volatile memory. The safety rectangle domain corresponding to the standard healthy operating conditions is limited to... It is between 1.2 and 1.8 and Within a closed region ranging from 0.04 to 0.07, the composite alarm boundary line is established by extending the safety rectangle outward by 1.5 times the tolerance radius. Specifically, the composite alarm boundary line passes through the geometric center of the safety rectangle. Perform linear scaling to establish the safety rectangle, and calculate the half-width of the safety rectangle in both dimensions. ,as well as Multiplying the half-width by the tolerance factor coefficient of 1.5, the boundary half-widths of the composite alarm area are obtained as follows: and The composite alarm boundary line constructed in this way forms a new rectangular discriminant surface in the two-dimensional topological phase space. The decision interval for the axis is [1.05, 1.95]. The axis determination interval is [0.0325, 0.0775], when the real-time two-dimensional data points When any coordinate value jumps out of the corresponding interval, the alarm state determination unit performs a logical XOR operation and determines that the abnormal state is established. At this time, the real-time feature point coordinate pair obtained by the alarm state determination unit is (2.49, 0.035). The coordinate trajectory of this dynamic topology node crosses the limit cutoff point of the composite alarm boundary line due to the electro-pneumatic impedance deflection, and the molecular sieve abnormal state is determined to be established. The alarm signal output unit includes a counter and a status register. When the counter detects that the dynamic topology node exceeds the composite alarm boundary line for 5 consecutive adsorption cycles, the abnormal state is finally confirmed. The status register inside the control system performs a five-level continuous shift confirmation and a state jump occurs, changing from the normal register value 00 to the fault alarm state 11. The alarm signal output unit then changes the output level of the main control chip pins, pulling the general-purpose input / output port high to generate a jump alarm signal and sending it to the audible and visual alarm response module. The audible and visual alarm response module, connected to the alarm signal output unit, controls the alarm red LED on the main control board to flash at a frequency of 3Hz and drives an external speaker to output an intermittent audio alarm with a decibel value of 105dB. This achieves feedforward graded alarm prompts for early-stage latent structural faults in the molecular sieve. During the above abnormal state determination process, the system also uses a leakage deviation compensation unit to offset the nonlinear deviation caused by mechanical leakage. The distribution valve experiences mechanical clearance wear and internal leakage in the gas path under long-term high-frequency switching impact, resulting in the lowest valley pressure value corresponding to the desorption period. Unable to reduce to the baseline safe negative pressure, the leakage deviation compensation unit continuously monitors the desorption valley pressure, and extracts the lowest valley pressure value. Compared to the historical resting negative pressure benchmark showing a monotonically increasing trend, a linear proportional operator is used to shift the first characteristic parameter in the two-dimensional virtual control topology phase space to the right in real time. The fault determination boundary line is thus adaptively calibrated to address the nonlinear interactive interference caused by the reduced local airtightness of the distribution valve components in determining the molecular sieve's degradation state. This represents the lowest valley pressure value, in units of... In actual compensation execution, the linear proportional operator uses a preset leakage mapping coefficient. Convert pressure drift into boundary displacement. The calculation formula is: ,in, The standard negative pressure reference value calibrated at the factory, and the mapping coefficient. The preset value is 12.5 based on the mapping curve between the rated diameter of the distribution valve and the exhaust back pressure. If it is detected... As wear increases from -0.05 MPa to -0.03 MPa, the boundary translation operator is calculated to... At this time, the alarm status determination unit automatically determines the composite alarm boundary line regarding... The upper limit of the positive discrimination of the axis was synchronously shifted from 1.95 to 2.20. Through this dynamic boundary envelope following mechanism, the system effectively offset the overall shift in the pressure build-up slope caused by internal leakage of the distribution valve.

[0049] Meanwhile, to verify the anti-false alarm capability of the alarm detection system of this invention, a comparative test under normal equipment power supply conditions was introduced. When the input voltage of the power supply network experiences an 8% transient voltage drop, the compressor's exhaust efficiency is limited due to insufficient input power of the drive energy field, and the built-up pressure slope of the adsorption tower slows down. At this time, the traditional system based on the static absolute pressure threshold directly triggers a false alarm at the lower pressure limit. However, when using the method of this technical solution, the voltage drop on the grid side causes oscillations of the harmonic reactive components inside the motor, and the second characteristic parameter calculated by the energy field characteristic conditioning unit... A nonlinear surge occurs, with the value jumping to 0.125, while the first characteristic parameter reflecting the separation impedance of the molecular sieve surface... The value is reduced to 0.85, and the characteristic parameter pair obtained by the alarm status determination unit becomes (0.85, 0.125). The alarm status determination unit retrieves the internal conditional control flow for causal comparison and determines that the point accurately falls within the low-frequency interference zone outside the power supply energy field and does not touch the fault determination red line of the composite alarm boundary. Based on this, the alarm signal output unit intercepts and suspends the system abnormal alarm signal and suspends the transmission of the system malfunction alarm signal. The instruction is downgraded to a yellow LED low-level power grid prompt. The bidirectional constraint of the information domain operator successfully offsets the external energy field noise interference. This technical solution achieves this within a single cycle of voltage swing adsorption. By establishing a cross-dimensional bidirectional dependency between the power supply side current load characteristics and the pneumatic pipeline pressure build-up slope, the alternating operating condition distortion at the physical level is transformed into dynamic topological node flow in phase space. Without changing the hardware base, the transformation from passive over-limit comparison at the end to active feedforward characteristic judgment of front-end impedance is realized. It not only provides a fault feedforward identification channel with causal logic chain support, but also constructs an adaptive hedging mechanism at the system level to cope with valve body wear and grid-side voltage fluctuations, so that the overall alarm control status output can stably converge to the actual physical structure damage of the bed under non-ideal service environment.

[0050] Example 2: This experiment aims to verify the anti-false alarm capability and real fault identification performance of the alarm detection system under complex operating conditions. The experiment was conducted on a simulated pneumatic test bench with pressure swing adsorption (PSA) cycle capability. The test object was a molecular sieve oxygen generator prototype equipped with the alarm detection system of this invention. The test bench has pressure and flow regulation capabilities. The pressure sensor sampling frequency is 50Hz, and the current sampling shunt sampling frequency is 1kHz to record the temporal fluctuations of the gas path and the power supply side. In the initial stage of the experiment, the system performed a 30-minute steady-state smoothing operation until the pneumatic pressure cycle data sequence obtained by the alarm state determination unit was in the statistically stable range, and the pressure reference value fluctuation range was controlled within 0.01MPa. The experiment was conducted in three groups: the present invention sample group, the control sample group, and the boundary sample group. In the test sequence of the present invention sample group, a high-frequency disturbance with a PSA cycle frequency of 0.1Hz was injected in real time, and a 5% leakage rate caused by the wear of the distribution valve was simulated. The pressure change parameters of this sample group were... The disturbance test procedure is as follows: The opening duty cycle of the distribution valve drive electromagnet is changed in real time through the PWM output of the microcontroller chip. Specifically, a high-frequency disturbance component with an amplitude of 10% of the rated voltage is superimposed on a standard 0.1Hz cyclic carrier wave, causing the distribution valve core to generate slight mechanical chatter during the commutation transient. This introduces a manually controlled frequency domain characteristic into the pressure waveform to test the signal extraction accuracy of the system under non-steady airflow impact. When the system enters the adsorption pressurization stage, the pneumatic circuit pressure increases with time at a slope of 0.52 MPa / s. The disturbance was detected in the second cycle. The value is 1.45, in the 5th period. The value was adjusted to 1.48, indicating that the system maintains stable pressure build-up characteristics within the leakage tolerance threshold. At this point, the second characteristic parameter... The fluctuation amplitude was maintained within 0.002. When a fault was introduced in the 10th cycle, the decrease in bed porosity was simulated by adjusting the pneumatic bypass, and the result was calculated. The value jumped to 2.12, 2.35, and 2.48 respectively within three consecutive cycles, and was simultaneously monitored. A decrease of 0.012 was observed, and the sample group of this invention successfully triggered the audible and visual alarm after 5 consecutive cycles of confirmation, with an alarm response delay of 0.25s.

[0051] The comparison sample design removes the current change parameter feedback logic of the current signal conditioning unit, relying solely on the first characteristic parameter. Alarm determination is based on a single dimension. Under the same fatigue conditions, because the influence of changes in the fundamental current envelope on the compressor load is ignored, when an 8% transient voltage drop occurs on the external grid side, the alarm detected in this group will be... Although the values ​​exhibit abnormal fluctuations due to the slowdown in pressure buildup, they lack... The system, through parameter-based collaborative arbitration, cannot distinguish whether the slowdown in voltage buildup is due to unstructured voltage fluctuations or a decrease in flow channel impedance caused by bed structure pulverization. This results in a false alarm occurring even under non-fault conditions. A value of 1.15 indicates that single-dimensional feature parameters have inherent recognition boundaries when dealing with electromagnetic interference. The boundary sample group design involves manually adjusting the radius of the safe rectangular domain boundary of the composite alarm boundary line to 0.5 times the original set value, thus placing it into the high-sensitivity false alarm range. During extreme boundary value testing, the degree of molecular sieve pulverization is set to only cause... A slight fluctuation of 0.15 was generated, which is within the normal operating fluctuation range. However, due to the setting of the allergy judgment boundary, the system triggered a fault alarm signal in the third pressurization cycle, indicating that when the judgment boundary deviates from the optimal working window setting, the system will lose its ability to suppress the fluctuation of the basic operating condition.

[0052] Based on the above test data, the alarm detection system of the present invention, through... and The dual-axis collaborative mapping, utilizing the physical correlation between gas flow impedance characteristics and changes in motor electromagnetic load, eliminates false alarms caused by single-physical-field identification. Data correlation analysis shows that... Between 2.0 and 2.5 and Within the characteristic parameter window of 0.03 to 0.04, the system exhibits optimal fault identification stability. The experimental results confirm that the alarm judgment mechanism based on multi-dimensional parameter fusion can accurately distinguish between external interference and internal physical structure damage, and the set composite alarm boundary line is in the optimal range that is feasible for engineering.

[0053] Example 3: This example combines Figures 1 to 2 This document describes an alarm detection system and method for a molecular sieve oxygen generator, such as... Figure 1 As shown, step S1 involves acquiring the pressure data sequence of the closed pressurization period of the pneumatic circuit within the molecular sieve pressure swing adsorption cycle through the pressure signal conditioning unit, calculating the first-order time difference value of the pressure data sequence, and outputting the pressure change parameter. Step S2 involves acquiring the current frequency signal driving the compressor through the current signal conditioning unit, extracting the rate of change of the fundamental amplitude envelope of the current frequency signal, and outputting the current change parameter. Step S3 involves receiving the pressure change parameter and the current change parameter through the alarm state determination unit, constructing two-dimensional data points characterizing the system's operating characteristics, and determining that the molecular sieve abnormal state is established when the two-dimensional data points exceed the set composite alarm boundary line. Step S4 involves changing the output level state of the main control chip pin through the alarm signal output unit when the number of consecutive cycles of the abnormal state reaches the set counting threshold, thereby generating a jump alarm signal.

[0054] like Figure 2As shown, the pressure signal conditioning unit collects the pressurization data sequence and outputs the changing parameters. The output of this conditioning unit is connected to the frequency domain filtering module and the alarm status determination unit via a dashed line. The leakage deviation compensation unit collects the lowest valley pressure value and shifts the boundary line, and is connected to the alarm status determination unit. The current signal conditioning unit extracts the envelope change rate and outputs the changing parameters, and is connected to the alarm status determination unit. The power supply monitoring unit collects the input voltage data and maintains the composite boundary, and is connected to the alarm status determination unit. The alarm status determination unit constructs two-dimensional data points to determine the abnormal state and connects to the alarm signal output unit at the next level. At the same time, the frequency domain filtering module calculates the high-frequency amplitude integral energy adjustment threshold and is also connected to the alarm signal output unit via a dashed line. The alarm signal output unit is responsible for outputting an alarm when the number of consecutive alarm cycles reaches the threshold, and is connected to the end-point audible and visual alarm response module. Finally, the audible and visual alarm response module executes diode flashing and the speaker outputs an alarm sound.

[0055] Example 4: In long-term service molecular sieve oxygen generators, sensor zero-point drift caused by ambient temperature fluctuations and the slow evolution of dynamic impedance due to deposits in the pneumatic pipeline can lead to unexpected shifts in the system alarm threshold. To ensure alarm accuracy during continuous operation, this example employs a dynamic baseline reconstruction strategy. Zero-point drift is automatically calibrated by sliding fit of historical state data. During the molecular sieve pressure swing adsorption cycle, the system acquires real-time pressure data sequences from the pneumatic circuit. And establish the associated pressure reference drift. The amount of drift The determination procedure is as follows: in each After one pressurization cycle, with this Adsorption tower reference pressure value within a cycle The mean sliding window operation is performed on the original data, and the specific calculation relationship is as follows: ,in, This is the pressure reference drift. For the first The baseline pressure value of the adsorption tower for each pressurization cycle. The sliding period window span is set to 1000, and the system calculates this value accordingly. For real-time pressure signals The corrected pressure signal is obtained by performing a subtraction operation. , using Substitute the above The calculation process eliminates the static error introduced by environmental temperature drift.

[0056] To further address the long-period evolution of the current fundamental amplitude envelope due to component aging, the system executes a secondary verification process based on drift compensation gain. When the electromagnetic load characteristics of the compressor motor deviate from the standard curve, the system automatically updates the second characteristic parameter. compensation value The calculation logic is as follows: ,in, The current characteristic parameter after compensation. The original feature parameters extracted by the current signal conditioning unit. This is the sensitivity coefficient, with a value of 0.15. It is the difference between the current real-time pressure reference drift and the initial pressure reference drift, i.e. .

[0057] At the 5000th cycle, the pressure signal conditioning unit detected wear on the valve body. The pressure is increased to 0.02 MPa, and the system performs dynamic offset calibration on the judgment boundary accordingly. Under this condition, the judgment unit extracts feature parameter pairs. The system projects this data into phase space. If the coordinate point remains within the composite alarm boundary line, the system maintains normal operation. If the feature pair crosses the composite alarm boundary line, the system switches the pressure swing adsorption frequency from 0.1Hz to a low-frequency pulse scanning mode of 0.08Hz to verify whether the deviation originates from the physical impedance degradation of the pneumatic circuit. In the low-frequency scanning mode, if the impedance response curve of the pneumatic circuit matches the preset standard pulverization model, it confirms that the molecular sieve bed has suffered physical structural damage and outputs a high-frequency audible and visual alarm. Through this logic link that introduces dynamic redundancy verification, effective defense against signal drift caused by equipment aging is achieved, ensuring the accuracy of the judgment of actual bed damage under complex operating conditions.

[0058] Example 5: In the commissioning scenario of initial deployment or replacement of the molecular sieve bed in a molecular sieve oxygen generator, due to differences in manufacturing processes, there is an objective deviation in the initial impedance of the gas path. The system executes an offline calibration program to establish the judgment benchmark for each state. Within 300 seconds after power-on, the system disables the real-time alarm logic for molecular sieve structural damage, forcibly opens the distribution valve, connects the adsorption tower to the external atmospheric environment, and collects 100 sets of pneumatic circuit pressure data. The arithmetic mean was calculated as the environmental reference pressure. The system is based on this reference pressure. Recalibrate the first characteristic parameter Security trigger threshold The specific calculation formula is as follows: ,in, The corrected alarm trigger threshold. The factory threshold value is 2.0 under standard sea level atmospheric pressure conditions. The current environmental reference pressure is automatically measured by the system. The standard atmospheric pressure at sea level is set at 0.1013 MPa. The aerodynamic drag correction factor is set to 0.65. After the pressure reference is reset, the current signal conditioning unit synchronously measures the leakage current of the motor coil in the resting state. And store it in an internal register as the second feature parameter. The static bias is set to eliminate static drift interference caused by sensor circuit aging. The correction coefficient k is set to 0.65, which is derived from the correlation fitting between the pressure slope and the Reynolds number of the ambient atmospheric pressure in the gas flow of porous media. Since the gas flow in the molecular sieve bed is in the transition zone between laminar and turbulent flow, according to the Elgun equation, the correlation between the pressure build-up slope and the gas density is not linear, but exhibits a power function characteristic. Under the narrow-diameter pipeline conditions of this system, comparative experiments at different altitudes showed that the ratio of the logarithmic change rate of the pressure response time constant to the logarithmic change rate of the gas pressure converges to around 0.65. Therefore, this coefficient is set to ensure the stability of the system under high-altitude environments. The correction can accurately restore the true flow resistance of the bed.

[0059] Once the oxygen concentrator enters a stable operating mode, the alarm status determination unit will call the calibrated value in real time. As a criterion, the real-time sampled value of the current signal conditioning unit will also be used. minus Differential processing is performed to obtain the corrected current characteristic parameters. If during continuous operation, if... Parameter touch The system triggers secondary logic verification by applying an excitation pulse with a frequency of 0.5Hz through the current characteristic conditioning unit and observing the impedance response characteristics of the pneumatic circuit. If the slope of the response curve deviates from the standard structural model by less than 5%, it is determined that the pressure fluctuation is caused by sensor zero-point drift due to changes in ambient temperature and humidity. The system automatically updates the offset compensation parameters. If the deviation exceeds 5%, it is confirmed that there is physical damage. The alarm signal output unit is controlled to drive the audible and visual alarm execution module to ensure that the judgment logic focuses only on the actual molecular sieve bed structure changes, rather than external environmental interference.

[0060] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. An alarm detection system for a molecular sieve oxygen generator, characterized in that, include: The pressure signal conditioning unit is connected to the alarm status determination unit, the current signal conditioning unit is connected to the alarm status determination unit, and the alarm status determination unit is connected to the alarm signal output unit. The pressure signal conditioning unit collects the pressure data sequence of the closed pressurization period of the pneumatic circuit in the molecular sieve pressure swing adsorption cycle, calculates the first-order time difference value of the pressure data sequence, and outputs the pressure change parameter. The current signal conditioning unit acquires the power frequency signal driving the compressor, extracts the rate of change of the fundamental amplitude envelope of the power frequency signal, and outputs the current change parameter. The alarm status determination unit receives pressure change parameters and current change parameters, constructs two-dimensional data points characterizing the system's operating features, and determines that the molecular sieve abnormal state is established when the two-dimensional data points exceed the set composite alarm boundary line; the alarm signal output unit changes the output level state of the main control chip pin when the number of consecutive cycles of the abnormal state reaches the set counting threshold, generating a jump alarm signal.

2. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, The system also includes a leakage deviation compensation unit; the leakage deviation compensation unit is connected to the alarm status determination unit, collects the lowest valley pressure value during the desorption period to determine the leakage value of the distribution valve, and converts the leakage value into a correction weight inside the alarm status determination unit. In each pressure swing adsorption cycle, the composite alarm boundary line is shifted according to the change in the correction weight.

3. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, The system also includes a frequency domain filtering module; the frequency domain filtering module is connected between the pressure signal conditioning unit and the alarm signal output unit, calculates the amplitude integral energy value of the pressure data sequence in the preset high frequency band, and adjusts the set counting threshold in the alarm signal output unit to the multi-sampling period accumulated value when the amplitude integral energy value is greater than the preset vibration threshold value.

4. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, The alarm signal output unit includes a counter and a status register connected to the counter. When the counter receives a composite alarm boundary line exceeding the judgment signal within 2405 consecutive pressure swing adsorption cycles, the abnormal state is finally confirmed, and the stored value of the status register is changed, driving the output level of the main control chip pin to switch to a level transition signal.

5. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, The pressure signal conditioning unit calculates the pressure change parameters by including the following steps: acquiring the pressure transient waveforms of the pneumatic circuit during the pressurization and stabilization phases, and discretizing the pressure transient waveforms into a pressure data sequence; calculating the discrete point difference of the pressure data sequence, and extracting the value representing the maximum slope of the pressurization process as the pressure change parameter.

6. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, The current signal conditioning unit extracts current change parameters by including the following steps: acquiring the fundamental component signal in the current power frequency signal; extracting the amplitude envelope of the fundamental component signal, and calculating the rate of change of the amplitude envelope with time as the current change parameter.

7. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, When the alarm signal output unit generates a jump alarm signal, it includes the following steps: at the end of each voltage swing adsorption cycle, the judgment result output by the alarm state judgment unit is accumulated to obtain the number of consecutive valid cycles; when the number of consecutive valid cycles reaches 2405 consecutive sampling cycles and all abnormal states are valid, the output level state of the main control chip pin is changed.

8. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, The system also includes an audible and visual alarm response module; the audible and visual alarm response module is connected to the alarm signal output unit and includes a light-emitting diode and a speaker; the light-emitting diode receives the jump alarm signal in an abnormal state and flashes at a frequency of 3Hz; the speaker outputs intermittent audio alarm sounds.

9. The alarm detection system for a molecular sieve oxygen generator according to claim 1, characterized in that, The system also includes a power supply monitoring unit; the power supply monitoring unit is connected to the alarm status determination unit and collects grid-side power supply input voltage data; when the power supply input voltage data is lower than the set power supply threshold and the pressure change parameter decreases monotonically, the alarm status determination unit keeps the composite alarm boundary line within the current set range.

10. A method for detecting an alarm in a molecular sieve oxygen generator, used in operating the alarm detection system for a molecular sieve oxygen generator as described in claim 1, characterized in that, include: Step S1: Collect the pressure data sequence of the closed pressurization period of the pneumatic loop in the molecular sieve pressure swing adsorption cycle through the pressure signal conditioning unit, calculate the first-order time difference value of the pressure data sequence, and output the pressure change parameter. Step S2: Acquire the power frequency signal of the current driving the compressor through the current signal conditioning unit, extract the rate of change of the fundamental amplitude envelope of the power frequency signal, and output the current change parameter. Step S3: The alarm status determination unit receives pressure change parameters and current change parameters, constructs two-dimensional data points characterizing the system operation characteristics, and determines that the molecular sieve abnormal state is established when the two-dimensional data points exceed the set composite alarm boundary line. Step S4: When the number of consecutive cycles of an abnormal state is established reaches a set counting threshold, the alarm signal output unit changes the output level of the main control chip pin to generate a jump alarm signal.

Images