Anti-interference type infrared human body induction air curtain machine starting and stopping device

Abstract

The invention relates to the technical field of automatic start-stop control, in particular to an anti-interference type infrared human body induction air curtain machine start-stop device, which realizes dynamic estimation and weighted smoothing of observation noise by introducing Kalman filtering into conjoint analysis of an energy peak value and an illumination gradient amplitude, improves signal stability under an illumination disturbance condition, and improves the stability of an infrared human body induction air curtain machine. The time-space consistency and the signal comparison precision of channel input are enhanced, the response proportion difference is calculated in a continuous period, and the posterior probability is updated by adopting Bayesian inference, so that the self-adaptive optimization of cross-period state recognition is achieved, the fluctuation of the human body existence probability is converged to a determinable interval, and the accuracy of human body state recognition is improved. According to the method, the feedback correction is constructed according to the energy proportion change and the state probability deviation, and the channel weight is dynamically adjusted, so that the judgment deviation caused by environment drift is reduced, the suppression capability of the system on baseline drift under the long-term operation condition is improved, and the response delay and the control stability are optimized.

Description

Technical Field

[0001] This invention relates to the field of automatic start-stop control technology, and in particular to an anti-interference infrared human body sensing air curtain machine start-stop device. Background Technology

[0002] The field of automatic start-stop control technology focuses on automatic control systems for equipment operation based on external environmental variables and target states. Specifically, it involves the detection of temperature, humidity, light, airflow, and other environmental factors, and the driving of controlled equipment to perform on / off, start-stop, and mode switching operations based on the detection signals. The technical solutions in this field typically include sensing units, signal processing units, control execution units, and feedback adjustment modules. The core objective is to achieve automated control of equipment operation, reduce intervention, lower energy consumption, and improve system response efficiency and environmental adaptability.

[0003] The anti-interference infrared human body sensing air curtain machine start-stop device is an intelligent control module installed in the control circuit of the air curtain machine. The device includes an infrared sensing unit, a spectral signal analysis unit, a data filtering and processing unit, and a relay drive control unit. The design purpose is to solve the technical problem of existing infrared sensors misinterpreting the presence of human body signals under strong light and direct sunlight. The device achieves high-precision judgment of the presence of human body status through multispectral signal discrimination and probability filtering calculation, thereby controlling the start-stop logic of the air curtain machine. The device can maintain a detection accuracy of 99.5% in complex lighting environments, achieving energy-saving and comfortable control effects such as automatic opening when a human body approaches and automatic closing when a human body leaves, reducing ineffective running time, reducing energy consumption, and extending the service life of the air curtain machine.

[0004] Existing technologies rely on a single infrared channel and fixed threshold logic, failing to establish a time-related structure between infrared energy signals and changes in illumination. Under strong light conditions, they are susceptible to false triggers caused by reflections and shadows. The signal processing flow lacks dynamic filtering and prediction correction, resulting in observation noise and transient fluctuations being directly amplified in the output, leading to judgment jitter and misjudgment of status. During the gradual transition phases of human movement, there is a lack of cross-cycle probability assessment mechanisms, resulting in delayed and oscillating system responses. The system cannot accurately distinguish between short-term occlusion and actual departure. During long-term operation, the slow drift and accumulation due to external temperature, light intensity, and sensor aging cause threshold judgment deviations, increasing false alarms and missed alarms, reducing energy consumption control and comfort adjustment effects, and resulting in insufficient overall anti-interference performance of the system. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and propose an anti-interference infrared human body sensing air curtain machine start-stop device.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: An anti-interference infrared human body sensing air curtain machine start / stop device includes: Spectrum acquisition module: It acquires signals of infrared detection voltage, visible light illuminance, temperature detection and time sampling through sensors, records synchronous readings of infrared and visible light channels, compares voltage amplitude difference with illuminance change direction, matches temperature increment with time interval, and constructs original channel response sequence. Signal determination module: Based on the original response sequence of the channel, extract the infrared energy peak, illumination gradient amplitude and heat source response interval, calculate the parameter difference and determine the interference exceeding the limit, introduce Kalman filtering to smooth the signal and correct the data, and generate the infrared effective response rate; State inference module: Based on the infrared effective response rate, calculate the continuous period response ratio difference and signal duration interval, compare the state transition trend with the threshold relationship, introduce Bayesian inference to update the state probability, and count the duration and occurrence of each state to obtain the human presence determination value. Dynamic correction module: Based on the infrared effective response rate and the human presence determination value, calculate the energy ratio change and state probability deviation, compare the deviation amplitude with the stable interval threshold, adjust the channel weight parameters, calculate the average rate of deviation difference, and output interference compensation parameters. The execution control module reads the direction and amplitude of the human body state change based on the interference compensation parameters, compares the amplitude value with the start and stop thresholds, outputs start and stop signals, calculates the time difference of the control signal and records the period, generates system commands, and constitutes start and stop trigger instructions.

[0007] As a further aspect of the present invention, the original response sequence of the channel includes synchronous readings of the infrared channel, records of visible light illuminance changes, temperature detection increments and time sampling indexes; the effective infrared response rate includes infrared energy peak value, illuminance gradient amplitude, heat source response interval and interference over-limit judgment parameters; the human presence judgment value includes continuous periodic response ratio difference, signal duration interval, state transition trend and probability update result; the interference compensation parameters include energy ratio change, state probability deviation, channel weight parameters and deviation difference average rate; and the start / stop trigger command includes the direction of human state change, change amplitude, comparison result of start-up threshold and stop-down threshold, and control signal time difference.

[0008] As a further aspect of the present invention, the spectral acquisition module includes: Channel acquisition submodule: Acquires infrared detection voltage signal, visible light illuminance signal, temperature detection signal and time sampling signal through sensors, performs synchronous reading and time recording, compares infrared voltage amplitude and illuminance direction, pairs temperature increment and time interval, and generates channel synchronous record value; Feature mapping submodule: Based on the channel synchronization record value, compare the difference between infrared and visible light amplitudes, compare the temperature rate by time index and locate the synchronization point, mark the light rise and fall intervals and sort the temperature increments to establish the original response sequence of the channel.

[0009] As a further aspect of the present invention, the signal determination module includes: Peak identification submodule: Based on the original response sequence of the channel, extract infrared energy readings and compare them with the light intensity curve. After locating the peak position through time series, calculate the amplitude difference, record the inter-peak spacing and energy position, determine and mark the peak start and end points, introduce Kalman filtering for sequence smoothing and peak position correction, and establish energy characteristic distribution values. Interference filtering submodule: Based on the energy characteristic distribution value, it detects the pulse amplitude and locates the reflected interference, determines the abnormal segment by comparing the deviation and deletes it, rearranges the remaining sequence according to the time index, marks the continuous interval and the direction of change and then checks the index to generate an effective signal sequence diagram; Stable statistics submodule: Based on the effective signal sequence diagram, identify continuous periodic infrared segments, calculate the proportion of stable segments in terms of duration and frequency of occurrence, summarize the time distribution and frequency count, form a periodic proportion and duration matching table, and generate the effective infrared response rate.

[0010] As a further aspect of the present invention, the Kalman filter performs dynamic estimation and correction of the time series based on the continuous infrared energy readings and light intensity curve change data in the original response sequence of the channel. It extracts the infrared energy values ​​of adjacent moments to calculate the prediction error and performs weighted smoothing on the deviation amplitude. It calculates the correction amount based on the difference between the energy state of the previous period and the current observed signal. The correction amount is superimposed on the current energy estimate to generate a smoothed sequence. Then, the peak position is repositioned in the smoothed sequence and the start and end intervals of the peak are adjusted. After the peak position correction is completed, the energy amplitude and spacing records are updated to form the corrected sequence data.

[0011] As a further aspect of the present invention, the state inference module includes: The period comparison submodule calculates the ratio difference of continuous detection periods based on the infrared effective response rate, extracts the response values ​​of adjacent periods and calculates the difference for comparison, records the direction of increase or decrease of the ratio and marks the time position, calculates the duration of each period interval, introduces Bayesian inference to evaluate the period probability, and organizes the time distribution to generate a period response difference table. Trend recognition submodule: Based on the periodic response difference table, it matches the direction of the proportional difference and the duration interval, determines the state change trend by distinguishing between rising and falling segments, counts the number of transitions in each period and records the position of the change segment, establishes a list of transition directions to integrate periodic information, and generates a state change trend graph. The results summary submodule: Based on the state change trend chart, it statistically analyzes the proportion and duration of multiple-cycle states, extracts the state occurrence frequency and average duration summary data, calculates the stability and transition ratios and records the change time points, integrates the continuous distribution and occurrence frequency, and obtains the human body presence determination value.

[0012] As a further aspect of the present invention, the Bayesian inference is based on the ratio difference and time position data in the infrared effective response rate and periodic response difference table. Prior probabilities are set and the observed values ​​are mapped. Then, a posterior update calculation is performed. The previous period state probability is read as the prior and jointly evaluated with the observation evidence corresponding to the current ratio difference and duration interval. The probability value of human presence in each period is obtained and written into the time index. Subsequently, the probability of adjacent periods is continuously updated and archived by combining the rising and falling segments in the state change trend diagram. A human presence determination value sequence synchronized with the periodic sequence is output.

[0013] As a further aspect of the present invention, the generation of the state change trend chart is based on the proportional difference direction data and duration interval data in the period response difference table output by the period comparison submodule. A state change sequence is established according to the time index order. The rising and falling directions are determined by identifying the proportional difference symbol and the corresponding period positions are marked. Adjacent periods with the same direction mark are continuously merged to form trend segments. The number of duration periods and the duration of each trend segment are calculated. The period number, proportional difference direction and segment duration are recorded as a trend segment table. An ordered data sequence containing period number, direction symbol and duration is generated by mapping the trend segment table according to the time index. The data sequence is mapped on the time axis to form a continuous curve of the state change direction, which constitutes a state change trend chart for the result summary submodule to perform multi-period state proportion and duration statistics.

[0014] As a further aspect of the present invention, the dynamic correction module includes: Deviation calculation submodule: Based on the infrared effective response rate and the human presence determination value, calculate the change in energy ratio, read and compare the two sets of period difference values, calculate the state probability deviation amplitude and compare it with the stable interval threshold, mark the deviation segment position, and generate channel deviation parameters. Weighting adjustment submodule: Based on the channel deviation parameters, determine the adjustment direction of the infrared and visible light channels, reallocate the channel ratio and correct the energy readings of each cycle, record the changed segments after calibrating the time index, calculate the average rate of the deviation difference and register it to obtain the interference compensation parameters.

[0015] As a further aspect of the present invention, the execution control module includes: Threshold judgment submodule: Based on the interference compensation parameters, read the direction and amplitude of human body state changes, mark the direction of continuous period energy difference and compare it with the start-up threshold and stop threshold, record the time point and amplitude of the over-limit section, and generate control judgment parameters; Signal output submodule: Based on the control judgment parameters, output start and stop signals, read the judgment direction to match the start and stop types and record the trigger time, calculate the time difference between the two signals to generate a control cycle record, and obtain the start and stop trigger command.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: 1. In this invention, by introducing Kalman filtering into the joint analysis of energy peak and illumination gradient amplitude, dynamic estimation and weighted smoothing of observation noise are achieved, which improves signal stability under illumination disturbance conditions and enhances the spatiotemporal consistency of channel input and signal comparison accuracy. 2. In this invention, by calculating the response ratio difference in continuous cycles and using Bayesian inference to update the posterior probability, adaptive optimization of cross-cycle state recognition is achieved, so that the fluctuation of the probability of human presence converges to the determinable range, thereby improving the continuity and accuracy of start and stop decisions. 3. In this invention, by constructing a feedback correction quantity based on changes in energy ratio and state probability deviation and dynamically adjusting channel weights, the judgment bias caused by environmental drift is reduced, the system's ability to suppress baseline drift under long-term operating conditions is improved, and the response delay and control stability are optimized. Attached Figure Description

[0017] Figure 1 This is a flowchart of the apparatus of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] Example 1 Please see Figure 1 This invention provides a technical solution: an anti-interference infrared human body sensing air curtain machine start / stop device comprising: Spectrum acquisition module: It acquires signals of infrared detection voltage, visible light illuminance, temperature detection and time sampling through sensors, records synchronous readings of infrared and visible light channels, compares voltage amplitude difference with illuminance change direction, matches temperature increment with time interval, and constructs original channel response sequence. Signal determination module: Based on the original response sequence of the channel, extract the infrared energy peak, illumination gradient amplitude and heat source response interval, calculate the parameter difference and determine the interference exceeding the limit, introduce Kalman filtering to smooth the signal and correct the data, and generate the infrared effective response rate; State inference module: Based on the infrared effective response rate, calculate the continuous period response ratio difference and signal duration interval, compare the state transition trend with the threshold relationship, introduce Bayesian inference to update the state probability, and count the duration and occurrence of each state to obtain the human presence determination value. Dynamic correction module: Based on the infrared effective response rate and the human presence determination value, calculate the energy ratio change and state probability deviation, compare the deviation amplitude with the stable interval threshold, adjust the channel weight parameters, calculate the average rate of deviation difference, and output interference compensation parameters. The execution control module reads the direction and amplitude of changes in human body state based on interference compensation parameters, compares the amplitude value with the start and stop thresholds, outputs start and stop signals, calculates the time difference of control signals and records the cycle, generates system commands, and constitutes start and stop trigger instructions.

[0020] The original response sequence of the channel includes synchronous readings of the infrared channel, records of visible light illuminance changes, temperature detection increments and time sampling indexes. The effective infrared response rate includes infrared energy peak value, light gradient amplitude, heat source response interval and interference over-limit judgment parameters. The human presence judgment value includes continuous periodic response ratio difference, signal duration interval, state transition trend and probability update results. The interference compensation parameters include energy ratio change, state probability deviation, channel weight parameters and deviation difference average rate. The start and stop trigger commands include the direction of human state change, change amplitude, comparison results of start and stop thresholds and control signal time difference.

[0021] The spectral acquisition module includes: Channel acquisition submodule: Acquires infrared detection voltage signal, visible light illuminance signal, temperature detection signal and time sampling signal through sensors, performs synchronous reading and time recording, compares infrared voltage amplitude and illuminance direction, pairs temperature increment and time interval, and generates channel synchronous record value; Feature mapping submodule: Based on the channel synchronization record value, compare the difference between infrared and visible light amplitudes, compare the temperature rate by time index and locate the synchronization point, mark the illumination rise and fall intervals and sort the temperature increments to establish the original response sequence of the channel; Channel acquisition submodule: Based on the infrared detection voltage signal, visible light illuminance signal, temperature detection signal and time sampling signal output by the sensor, the data acquisition is aligned with the time sequence. The sampling frequency is set to 100 times per second, the infrared voltage range is 0 to 5 volts, the illuminance range is 0 to 10000 lux, the temperature detection range is 0 to 60 degrees Celsius, and the time resolution is 0.01 seconds. Synchronous sampling is triggered by the master clock signal. The acquired data is recorded in a unified matrix according to infrared voltage, illuminance reading, temperature value and time mark. The sampling interval error is limited to ±0.002 seconds. The direction of illuminance change is determined by the difference of continuous sampling points and the rising and falling segments are marked. The temperature increment is taken as an average of 0.05 degrees Celsius and the time interval is taken as 0.01 seconds. After pairing, a unified time index matrix is ​​formed, and finally the channel synchronous record value is generated. Feature mapping submodule: Based on the channel synchronization record value, calculate the amplitude difference between infrared and visible light, convert the signal to the 0 to 1 interval, set the amplitude difference threshold to 0.15 and mark significant response segments, calculate the temperature change rate according to the time difference, set the range to 0.1 to 0.5 degrees Celsius per second, determine the synchronization point by the time correspondence between the rate and the amplitude difference, distinguish the rising and falling light change segments according to the degree direction label, and record them with the symbols 1 and -1, sort the temperature increments by size and pair them with the time index, and merge the infrared amplitude difference, light segment label, temperature rate and time index to form the original response sequence of the channel.

[0022] The signal determination module includes: Peak identification submodule: Based on the original response sequence of the channel, extract infrared energy readings and compare them with the light intensity curve. After locating the peak position through time series, calculate the amplitude difference, record the inter-peak spacing and energy position, determine and mark the peak start and end points, introduce Kalman filtering for sequence smoothing and peak position correction, and establish energy characteristic distribution values. Interference filtering submodule: Based on energy characteristic distribution values, it detects pulse amplitude and locates reflected interference, determines abnormal segments by comparing deviations and deletes them, rearranges the remaining sequence according to time index, marks continuous intervals and change directions, and then checks the index to generate an effective signal sequence diagram; Stable statistics submodule: Based on the effective signal sequence diagram, it identifies continuous periodic infrared segments, calculates the proportion of stable segments based on duration and frequency of occurrence, summarizes the time distribution and frequency count, forms a periodic proportion and duration matching table, and generates the infrared effective response rate; Peak Identification Submodule: Based on the original response sequence of the channel, the peak value of the infrared energy reading is extracted. The sampling interval is set to 0.01 seconds, and the infrared energy range is 0 to 5 volts. Continuous sampling points are compared point by point. When the current sampling value is greater than the previous and next sampling values, it is determined to be a local peak. The peak time and energy value are recorded in an index table. Then, the illumination intensity curve is smoothed. The window width is set to 5 sampling points, and the window step size is 1 sampling point. After calculating the smoothed curve, the infrared peak position is compared to determine the illumination time position corresponding to the peak position. The amplitude difference of the located peak point is calculated. The amplitude difference threshold is set to 0.2 volts. The inter-peak spacing is recorded and the time interval value is stored in seconds. The energy position is recorded by sampling point number. Kalman filtering algorithm is introduced for time series smoothing. The state transition equation adopts a linear prediction model. The measurement noise variance is set to 0.04, and the process noise variance is set to 0.01. The state estimate is updated recursively and the correction gain is calculated. Smoothing correction and peak position adjustment are performed on the energy value at each time moment to generate energy characteristic distribution values. Interference filtering submodule: Based on energy characteristic distribution values, it detects amplitude anomalies, sets the pulse threshold to 1.5 times the average energy of the current sequence, traverses each energy point, and marks pulse interference if the amplitude exceeds the threshold. Then, it calculates the average energy difference between adjacent intervals. If the deviation exceeds 0.3 volts, it is judged as a reflection interference segment and marked for deletion. The remaining sequence is renumbered with time index, with the time interval controlled at 0.01 seconds. The data is rearranged in chronological order. The direction of change of continuous intervals is determined by the positive and negative values ​​of adjacent energy differences. The direction is identified by 1 and -1 respectively. The rearranged sequence is stored in a table structure with fields including time index, energy amplitude, direction identifier and interval number. The index is checked and a valid signal sequence diagram is generated. Stable statistics submodule: Based on the effective signal sequence diagram, it identifies continuous infrared segments, sets the periodic sampling length to 100 sampling points, and determines the continuous periodic infrared segment as having an energy amplitude greater than the average value of 0.8 volts and a continuous sampling number of more than 20 points. It calculates the duration of each segment and records it in seconds, counts the occurrence frequency and calculates the proportion of stable segments. The time distribution record is indexed by the start time of each segment, and the frequency count is stored as an integer. It is organized in chronological order to form a periodic proportion table and a continuous matching table, and generates the infrared effective response rate.

[0023] Kalman filtering, based on the continuous infrared energy readings and light intensity curve changes in the original response sequence of the channel, performs dynamic estimation and correction of the time series. It extracts the infrared energy values ​​of adjacent time moments to calculate the prediction error and performs weighted smoothing on the deviation amplitude. It calculates the correction amount based on the difference between the energy state of the previous period and the current observed signal. The correction amount is superimposed on the current energy estimate to generate a smoothed sequence. Then, the peak position is repositioned in the smoothed sequence and the start and end intervals of the peak are adjusted. After the peak position correction is completed, the energy amplitude and spacing records are updated to form the corrected sequence data. Kalman filtering, according to the formula:

[0024] in: For the first Infrared energy state estimate at time step. For the first Infrared energy estimate at time. For the first Time-state covariance The process noise baseline variance, To measure the noise baseline variance, Forgetting factor, For the first Light intensity gradient over time, For the first Increment of illumination reading over time For the first Temperature increment over time For the first Infrared signal saturation ratio at any time This is the illumination gradient correction factor. This is the illumination gradient amplification factor. This is the amplification factor for the increase in illumination. The temperature squared weighting coefficient is used. This is the saturation ratio amplification factor. For the first Real-time infrared sensor measurement value; Execution process: First, estimate the state based on the previous time step. Covariance Predict the current state using parameters Adjust historical weights to reduce the impact of accumulated errors on new sampling, and calculate the illumination gradient. Increase in light intensity Temperature increment Compared with infrared saturation ratio Substitute the parameters into the adaptive measurement matrix correction term in the formula. With noise terms The measurement noise and process noise are corrected through a proportional relationship. The calculated ratio is the adaptive Kalman gain, used to balance the weights between the predicted and measured values. The device iteratively updates the energy state estimate at each sampling time using a formula, gradually forming a time-continuous smooth sequence, and then updates the result... Stored in a timing buffer and recorded synchronously with the light and temperature channel data.

[0025] The state inference module includes: The period comparison submodule calculates the ratio difference of continuous detection periods based on the infrared effective response rate, extracts the response values ​​of adjacent periods and calculates the difference for comparison, records the direction of increase or decrease of the ratio and marks the time position, calculates the duration interval of each period, introduces Bayesian inference to evaluate the period probability, and organizes the time distribution to generate a period response difference table. Trend recognition submodule: Based on the periodic response difference table, it matches the direction of the proportional difference and the duration interval, determines the state change trend by distinguishing between rising and falling segments, counts the number of transitions in each period and records the location of the change segment, establishes a list of transition directions to integrate periodic information, and generates a state change trend graph. The results summary submodule: Based on the state change trend chart, it statistically analyzes the proportion and duration of multi-cycle states, extracts the state occurrence frequency and average duration summary data, calculates the stability and transition ratio and records the change time points, integrates the continuous distribution and occurrence frequency, and obtains the human body presence determination value. The period comparison submodule calculates the proportional difference between consecutive detection periods based on the infrared effective responsivity. The detection period length is set to 1 second, with each period containing 100 sampling points. A step-by-step difference calculation is performed on the response values ​​of adjacent periods, with a difference threshold set to 0.05. Periods exceeding the threshold are marked as changing periods, and the changing time index and corresponding response value are recorded. Subsequently, a local average is performed on the proportional difference sequence, with a window width of 3 periods and a step size of 1 period. The average proportional difference within each window is calculated by accumulation and the results are stored. The proportional differences are then compared after extracting the response values ​​of adjacent periods, and the comparison results are recorded using direction indicators. When the difference is positive, it is marked as 1; when it is negative, it is marked as -1. The time position is recorded in the index table in seconds. When calculating the duration of each period interval, the interval value is obtained by subtracting the timestamps of adjacent periods. The interval value and the proportional difference are stored in the time table. Then, a Bayesian inference algorithm is introduced to evaluate the period probability. The prior probability distribution is set to uniform distribution with a prior value of 0.5. The input of the likelihood function is the absolute value of the proportional difference. The noise variance is set to 0.02. The posterior probability is recursively calculated using the Bayesian update formula, and the update value for each period is recorded. The probability value and the correspondence between time are organized into an ordered data table to generate a period response difference table. Trend Recognition Submodule: Based on the periodic response difference table, it matches the direction of the proportional difference and the duration interval. The input parameters are the proportional difference direction identifier and the time interval sequence. It judges the consistency of the direction of adjacent periods. If the continuous direction identifiers are the same and the duration is greater than 2 seconds, they are classified into the same trend segment. The rising and falling segments are classified separately. The classification number is recorded as rising segment 1 and falling segment -1. After the segment division is completed, the number of periods contained in each trend segment is calculated using the segment statistics method. The time positions of the start and end periods are recorded in the trend table. When counting the number of transitions in each period, the points of direction change are counted. The change direction is recorded as two categories: rising to falling and falling to rising. The position of the change segment is written into the index table in seconds. The period information is integrated into a transfer direction list in ascending order. The three data items of proportional difference direction, segment length and number of transitions are summarized to form a structured data record and generate a state change trend chart. The results summary submodule, based on the state change trend chart, statistically analyzes the proportion and duration of multiple-cycle states. The total number of cycles is set to 300, with a 1-second time interval between each cycle. The duration of the rising and falling segments in the trend chart is recorded for each cycle, and the total time is calculated in seconds. The average duration of the rising segment is 6 seconds, and the average duration of the falling segment is 4 seconds. The frequency of each state is extracted and recorded as an integer count, with a frequency range of 0 to 50 times. An average calculation is performed on the cycle data to obtain an average duration of 5 seconds. The ratio of stable states to transition states is calculated, with a stable state ratio of 0.65 and a transition state ratio of 0.35. The results are stored in the results table, rounded to two decimal places. When recording change time points, the direction switching index in the trend chart is used as a reference. The switching time is recorded in timestamp format in seconds. The duration distribution and occurrence frequency are integrated in a table, with fields including state type, duration, frequency of occurrence, transition ratio, and time index. The data is sorted in ascending order by time to obtain the human presence determination value.

[0026] Bayesian inference, based on the ratio difference and time position data in the infrared effective response rate and periodic response difference table, sets a prior probability and maps the observed values, then performs posterior update calculation, reads the previous period state probability as a prior and performs joint evaluation with the observation evidence corresponding to the current ratio difference and duration interval, obtains the human presence probability value for each period and writes it into the time index, then combines the rising and falling segments in the state change trend map to continuously update and archive the probabilities of adjacent periods, and outputs a human presence determination value sequence synchronized with the periodic sequence; Bayesian inference, according to the formula:

[0027] in: For the first The posterior probability of the presence of a human body in a given testing period. For the first Prior probability for each detection cycle For the first Average infrared energy over one detection cycle For the first Average ambient temperature over each testing period The expected value of the infrared signal in the presence of a human body. The standard deviation of infrared signals in the presence of the human body. The expected value of the infrared signal in an unmanned state. The standard deviation of the infrared signal in an unmanned state. The desired ambient temperature. For the standard deviation of ambient temperature, For the first Light interference suppression coefficient for each detection cycle For the first Signal stability parameters for each detection cycle For the first Noise compensation item for each detection cycle, For exponential function operators; Execution process: First, the signals of each detection cycle are sampled in the infrared detection and temperature channel, and the average infrared energy is calculated. With temperature average Then, the light interference suppression coefficient is determined by the rate of change of the light signal. The stability parameter is obtained by comparing the standard deviations of three consecutive infrared signals. The compensation term is determined based on the sampling noise level. Then, the parameters are substituted into the formula for calculation, where the exponents in the numerator and denominator are... The probability density distribution function represents the state with and without human presence. The negative sign in the exponential term is used to suppress signal fluctuations that deviate from the expected value, making them more responsive to human behavior. and The energy sample is weakened in probability, and the system completes one calculation in each cycle and obtains the posterior probability. .

[0028] A state change trend chart is generated based on the proportional difference direction data and duration interval data in the period response difference table output by the period comparison submodule. A state change sequence is established according to the time index order. The upward and downward directions are determined by identifying the proportional difference sign and the corresponding period positions are marked. Adjacent periods with the same direction mark are continuously merged to form trend segments. The number of duration periods and the duration of each trend segment are calculated. The period number, proportional difference direction and segment duration are recorded as a trend segment table. Based on the trend segment table, an ordered data sequence containing period number, direction sign and duration is generated by mapping according to the time index. The data sequence is mapped on the time axis to form a continuous curve of state change direction, which constitutes a state change trend chart for the result summary submodule to perform multi-period state proportion and duration statistics.

[0029] The dynamic correction module includes: Deviation Calculation Submodule: Based on the infrared effective response rate and the human presence determination value, calculate the change in energy ratio, read and compare two sets of period difference values, calculate the state probability deviation amplitude and compare it with the stable interval threshold, mark the deviation segment position, and generate channel deviation parameters. The weight adjustment submodule determines the adjustment direction of the infrared and visible light channels based on the channel deviation parameters, reallocates the channel ratio and corrects the energy readings of each cycle, records the changed sections after calibrating the time index, calculates and registers the average rate of the deviation difference, and obtains the interference compensation parameters. Deviation Calculation Submodule: Based on the infrared effective response rate and the human presence detection value, it calculates the change in energy ratio. The infrared effective response rate and the human presence detection value are both set to a range of 0 to 1, with a time sampling period of 1 second. A differential operation is performed on the infrared effective response rate for two consecutive periods, calculating the change in energy ratio for each period and recording the direction of change. Positive changes are marked as 1, and negative changes as -1. The differences between the two sets of periods are read and compared. A difference threshold of 0.05 is set; when the absolute difference between the two sets of differences exceeds the threshold, it is marked as abnormal. The system then uses a weighted average method to calculate the state probability deviation amplitude, with weighting coefficients set to 0.6 and 0.4. The infrared channel and the illumination channel are weighted separately, and the weighted deviation is calculated and compared with the stable interval threshold, which is set to 0.1. When the deviation amplitude exceeds the threshold, the start and end times of the deviation segment and the sampling point index are recorded. The deviation segment position numbers are recorded in ascending order. All deviation results are stored in matrix form, and the data fields include the period number, energy ratio change, deviation amplitude, threshold state, and time index, generating channel deviation parameters. The weight adjustment submodule determines the adjustment direction of the infrared and visible light channels based on the channel deviation parameters. It standardizes the energy ratio change in the channel deviation parameters, setting the initial weight of the infrared channel to 0.5 and the initial weight of the visible light channel to 0.5. When the deviation amplitude is positive and the infrared channel deviation is greater than 0.05, the infrared channel weight is increased to 0.6 and the visible light channel weight is decreased to 0.4. When the deviation is negative and the visible light channel deviation is greater than 0.05, the visible light channel weight is increased to 0.6 and the infrared channel weight is decreased to 0.4. The channel ratios are redistributed after weight adjustment, with the sum of channel weights remaining at 1. When correcting energy readings in each cycle, the energy value is recalculated using the adjusted weights. After calibrating the time index, the changed segment number and adjustment direction are recorded. Subsequently, the average rate of the deviation difference is calculated using a moving average method. The sliding window width is set to 3 cycles, and the step size is 1 cycle. The average rate value for each window is calculated and stored in a registration table. The registration table includes the time index, infrared channel weight, illumination channel weight, average rate, and segment number, thus obtaining the interference compensation parameters.

[0030] The execution control module includes: Threshold judgment submodule: Based on interference compensation parameters, read the direction and amplitude of human body state changes, mark the direction of continuous period energy difference and compare it with start-up threshold and stop threshold, record the time point and amplitude of the over-limit section, and generate control judgment parameters; Signal output submodule: Based on control judgment parameters, output start and stop signals, read the judgment direction to match the start and stop types and record the trigger time, calculate the time difference between two signals to generate control cycle records, and obtain start and stop trigger commands; Threshold Judgment Submodule: Based on interference compensation parameters, it reads and analyzes the direction and amplitude of changes in human body state. Input parameters include energy difference sequence, time index, and direction identifier. The sampling period is set to 1 second, and the energy amplitude range is set to 0 to 5 volts. Direction identification is performed on the difference in each cycle, with positive direction marked as 1 and negative direction marked as -1. Direction data and time index are stored in pairs. The start threshold is set to 3.2 volts, and the stop threshold is set to 1.8 volts. The energy difference is compared cycle by cycle. When the difference is greater than the start threshold or less than the stop threshold, a trigger mark is set. The error tolerance is set to ±0.02 volts. Repeated sampling verification is performed on boundary values. The number of consecutive samples is 3, and the maximum deviation between samples does not exceed 0.05 volts. A logical filtering algorithm is used to determine the continuity of the over-limit section. The threshold for the number of consecutive cycles is 3. The duration and average amplitude value of each over-limit section are counted. The standard deviation of the average amplitude is controlled within 0.1 volts. Abnormal sections are filtered out by the condition that the error variance does not exceed 0.004, and control judgment parameters are generated. Signal output submodule: Based on control judgment parameters, output start / stop signals. Input parameters include direction identifier, trigger time, amplitude value, and cycle number. A direction identifier of 1 corresponds to a start signal, and a direction identifier of -1 corresponds to a stop signal. Trigger types are marked as type A and type B signals. The trigger time recording accuracy is 0.001 seconds. The time difference between two consecutive signals is calculated and recorded, with the time difference range limited to 0 to 10 seconds. Error detection is performed on the trigger cycle, and the standard deviation of the cycle is controlled within 0.05 seconds. When the time difference exceeds the average value ±0.1 seconds, the invalid state is remarked and resampled. The output pulse width is controlled, with the pulse width set to 100 milliseconds. The output level is a 5-volt DC signal. The trigger error detection rate is controlled at 0.5%, meaning that the number of false triggers does not exceed 5 out of 1000 signal outputs. Output port number D2 executes the electrical signal transmission and records the transmission time. All trigger data is recorded in the control cycle table, and finally, the start / stop trigger command is obtained. The comprehensive detection accuracy of cycle synchronization error and amplitude difference error is maintained at 99.5%.

[0031] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. Anti-interference type infrared human body sensing wind curtain machine start-stop device, characterized in that, The device comprises: The spectrum acquisition module: through the sensor, infrared detection voltage, visible light illumination, temperature detection and time sampling signals are acquired, infrared and visible light channel synchronous readings are recorded, voltage amplitude difference and illumination change direction are compared, temperature increment and time interval are matched, and channel original response sequence is constructed; The signal determination module: based on the channel original response sequence, infrared energy peak value, illumination gradient amplitude and heat source response interval are extracted, parameter difference is calculated and interference overrun is determined, Kalman filter is introduced for signal smoothing and data correction, and infrared effective response rate is generated; The state inference module: based on the infrared effective response rate, continuous period response proportion difference and signal duration interval are calculated, state transition trend and threshold relationship are compared, Bayesian inference is introduced for state probability update, state duration and occurrence frequency are counted, and human body existence determination value is obtained; The dynamic correction module: based on the infrared effective response rate and the human body existence determination value, energy proportion change and state probability deviation are calculated, deviation amplitude and stable interval threshold are compared, channel weight parameter is adjusted, deviation difference average rate is calculated, and interference compensation parameter is outputted; The execution control module: based on the interference compensation parameter, human body state change direction and amplitude are read, amplitude value and start and stop threshold are compared, start and stop signals are outputted, control signal time difference is calculated and period is recorded, system command is generated, and start and stop trigger instruction is constituted.

2. The anti-interference type infrared human body sensing wind screen machine start-stop device according to claim 1, characterized in that, The channel original response sequence comprises infrared channel synchronous readings, visible light illumination change record, temperature detection increment and time sampling index, the infrared effective response rate comprises infrared energy peak value, illumination gradient amplitude, heat source response interval and interference overrun determination parameter, the human body existence determination value comprises continuous period response proportion difference, signal duration interval, state transition trend and probability update result, the interference compensation parameter comprises energy proportion change, state probability deviation, channel weight parameter and deviation difference average rate, and the start and stop trigger instruction comprises human body state change direction, change amplitude, start threshold and stop threshold comparison result and control signal time difference.

3. The anti-interference infrared human body sensing wind curtain machine start-stop device according to claim 1, characterized in that, The spectrum acquisition module comprises: The channel acquisition submodule: through the sensor, infrared detection voltage signal, visible light illumination signal, temperature detection signal and time sampling signal are acquired, synchronous reading and time recording are carried out, infrared voltage amplitude and illumination direction are compared, temperature increment and time interval are matched, and channel synchronous record value is generated; The feature mapping submodule: based on the channel synchronous record value, infrared and visible light amplitude difference are compared, temperature rate is compared according to time index and synchronous point is located, temperature increment is sorted after illumination rising and falling interval is marked, and channel original response sequence is established.

4. The anti-interference type infrared human body sensing wind screen machine start-stop device according to claim 1, characterized in that, The signal determination module comprises: The peak identification submodule: based on the channel original response sequence, infrared energy reading is extracted and illumination intensity curve is compared, amplitude difference is calculated after peak position is located through time sequence, peak interval and energy position are recorded, peak start and end points are determined and marked, Kalman filter is introduced for sequence smoothing and peak correction, and energy feature distribution value is established. The interference filtering submodule: based on the energy feature distribution value, positioning the reflected interference after detecting the pulse amplitude, determining the abnormal section by comparing the deviation, rearranging the remaining sequence by time index, marking the continuous interval and the change direction to correct the index, and generating the effective signal sequence graph; The stable statistics submodule: based on the effective signal sequence graph, identifying the continuous period infrared section, calculating the duration and frequency of the stable section proportion, summarizing the time distribution and frequency, forming the period proportion and duration matching table, and generating the infrared effective response rate.

5. The anti-interference infrared human body sensing wind curtain machine start-stop device according to claim 1, characterized in that, The Kalman filter, based on the continuous infrared energy readings in the channel original response sequence and the illumination intensity curve change data, performs dynamic estimation and correction of time series, extracts the infrared energy value at adjacent time to calculate the prediction error and weight the deviation amplitude for smoothing, calculates the correction amount based on the energy state of the previous period and the difference between the current observation signal, adds the correction amount to the current energy estimation value to generate a smoothed sequence, then repositions the peak position in the smoothed sequence and adjusts the peak start and end interval, updates the energy amplitude and interval record after completing the peak correction, and forms the corrected sequence data.

6. The anti-interference type infrared human body sensing wind screen machine start-stop device according to claim 1, characterized in that, The state inference module includes: The period comparison submodule: based on the infrared effective response rate, calculate the proportion difference of consecutive detection periods, extract the response values of adjacent periods and calculate the difference for comparison, record the proportion increase and decrease direction and mark the time position, calculate the length of each period interval, introduce Bayesian inference for period probability evaluation, and generate a period response difference table by organizing the time distribution; The trend identification submodule: based on the period response difference table, match the proportion difference direction and the duration interval, determine the state change trend by distinguishing the rising and falling sections, count the number of state transitions in each period and record the change section position, establish a transition direction list to integrate period information, and generate a state change trend graph; The result summary submodule: based on the state change trend graph, calculate the proportion and duration of multiple periods, extract the state occurrence frequency and average duration period, calculate the stable and conversion proportion and record the change time point, integrate the duration distribution and occurrence frequency, and obtain the human presence determination value.

7. The anti-interference infrared human body sensing wind curtain machine start-stop device according to claim 1, characterized in that, The Bayesian inference, based on the infrared effective response rate, the proportion difference and time position data in the period response difference table, sets the prior probability and maps the observation value, then performs posterior update calculation, reads the state probability of the previous period as the prior, and jointly evaluates the observation evidence corresponding to the current proportion difference and duration interval, obtains the human presence probability value of each period and writes it into the time index, then updates and archives the adjacent period probability in combination with the rising section and falling section in the state change trend graph, and outputs the human presence determination value sequence synchronized with the period sequence.

8. The anti-interference type infrared human body sensing wind curtain machine start-stop device according to claim 6, characterized in that, The generation of the state change trend chart is based on the proportional difference direction data and the continuous interval data in the period response difference table output by the period comparison submodule, a state change sequence is established in time index order, the rising and falling directions are determined by identifying the proportional difference signs and the corresponding period positions are marked, adjacent periods with the same direction mark are continuously merged to form trend sections, and the number of continuous periods and the time length of each trend section are calculated, the period number, the proportional difference direction and the section duration are recorded as a trend section table, an ordered data sequence containing the period number, the direction symbol and the duration is generated according to the trend section table in time index mapping, a continuous curve of the state change direction is formed by corresponding mapping of the data sequence on the time axis, and the state change trend chart is constituted for the result summary submodule to perform multi-period state proportion and duration statistics.

9. The anti-interference infrared human sensing wind curtain machine start-stop device according to claim 1, characterized in that, The dynamic correction module comprises: a deviation calculation submodule: based on the infrared effective response rate and the human body existence determination value, the energy proportion change amount is calculated, the two groups of period difference values are read and compared, the state probability deviation amplitude is calculated and compared with the stable interval threshold value, the deviation section position is marked, and the channel deviation parameter is generated; a weight adjustment submodule: based on the channel deviation parameter, the infrared and visible light channel adjustment direction is determined, the channel proportion is redistributed and each period energy reading is corrected, the changed section is recorded after time index correction, the deviation difference average rate is calculated and recorded, and the interference compensation parameter is obtained.

10. The anti-interference infrared human sensing wind curtain machine start-stop device according to claim 1, characterized in that, The execution control module comprises: a threshold judgment submodule: based on the interference compensation parameter, the human body state change direction and amplitude are read, the continuous period energy difference direction is marked and compared with the start threshold and the stop threshold, the out-of-limit section time point and amplitude are recorded, the control determination parameter is generated; a signal output submodule: based on the control determination parameter, the start and stop signal output is performed, the determination direction matching start and the type of closing are read and the trigger time is recorded, the control period record is generated by calculating the time difference of two signals, and the start and stop trigger instruction is obtained.

Images