A reducer control system based on chance constraint

Abstract

The application relates to the technical field of digital processors, in particular to a reducer control system based on an opportunity constraint, which comprises a dynamic window module, a temperature cooling module, a lubricating pressure module, a gear meshing module and an optimization control module.In the application, reducer torque and rotating speed data are collected, the difference value of adjacent time points is calculated, the state transition probability is estimated in combination with a Markov chain model, a transition time window and an error range are set, the control parameters are dynamically adjusted, the reducer state change is fed back in real time, the real-time performance and adaptability of the control system are improved, the temperature data interval matching and trend judgment avoid over-temperature failure, the stability is ensured, the lubricating oil pressure and the cooling correction result are combined and compared, the lubrication is ensured to be appropriate, wear and failure are avoided, gear vibration and temperature data are analyzed through a probability density function, the gear meshing state is adjusted, and the system robustness and operation efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of digital processor technology, and in particular to a speed reducer control system based on chance constraints. Background Technology

[0002] The field of digital processor technology encompasses core components that perform logical and numerical operations on input data and their applications. Its core content lies in achieving automated management of external devices or processes through program control. Digital processors typically possess functions such as instruction fetching, execution, data storage, and output, and are widely used in industrial control, embedded systems, communication equipment, and intelligent manufacturing.

[0003] Among them, the chance-constraint-based reducer control system refers to the design of its control logic by incorporating chance-constraint optimization theory during the reducer's operation. This involves introducing chance constraints into the control instruction set of the digital processor, handling the relationship between the reducer input and load through constraint modeling, and generating control parameters using optimization methods. Specifically, this includes setting parameters based on probabilistic constraints, determining constraints on input variables during the instruction execution phase of the digital processor, and applying the results to the reducer's operating logic through the program control unit, thus forming a control system scheme with chance-constraint characteristics.

[0004] In existing technologies, digital processor control systems largely rely on fixed control parameters and threshold settings, making dynamic adjustments based on real-time data impossible. During gearbox operation, traditional control schemes typically depend on fixed parameters and response rules, which struggle to handle complex and changing operating conditions, easily leading to control response lag and impacting the overall system's stability and reliability. Particularly when load fluctuations are large or external conditions such as temperature change rapidly, existing systems fail to provide real-time, precise feedback adjustments, potentially causing gearbox overheating, improper lubrication, and reduced system efficiency and lifespan. Furthermore, existing technologies fail to effectively consider the synergistic effects between various control components, making it impossible to comprehensively assess the impact of various factors on gearbox operation, easily resulting in control blind spots or inaccurate parameter adjustments. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and propose a speed reducer control system based on chance constraints.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a speed reducer control system based on chance constraints includes:

[0007] The dynamic window module collects the torque and speed data of the reducer and calculates the difference between adjacent time points. It estimates the state transition probability through a Markov chain model, sets the transition time window and error range, generates a dynamic transition window, and transmits it to the temperature cooling module.

[0008] The temperature cooling module acquires the gearbox temperature data based on the dynamic transition window and performs interval matching, calculates the temperature change trend and determines whether the temperature safety operating threshold is exceeded. If it is exceeded, the cooling power is increased, a cooling correction result is generated and transmitted to the lubrication pressure module.

[0009] The lubrication pressure module acquires lubricating oil pressure data and the current speed of the reducer, and performs a joint comparison with the cooling correction result. When the joint comparison result does not reach the lubricating oil pressure threshold, the lubrication pump supply is increased, a lubrication supply adjustment command is generated, and the command is transmitted to the gear meshing module.

[0010] The gear meshing module, based on the lubrication supply adjustment command, acquires the adjusted gear vibration data and gear temperature data and calculates the error respectively, inputs the probability density function to determine the probability centroid offset, adjusts the gear meshing state and gear clearance in the reducer, and generates the reducer control result;

[0011] The optimization control module, based on the reducer control results, acquires the adjusted short-time sampled torque and speed data for distribution statistics, corrects the probability centroid of the reducer control results by offset, and updates the dynamic transition window to generate the optimized control results for the reducer.

[0012] As a further aspect of the present invention, the dynamic transition window includes state transition probability, transition time interval, and error allowable range; the cooling correction result includes cooling power, temperature threshold difference, and temperature change rate; the lubrication supply adjustment command includes pump start / stop status and pressure compensation amount; the reducer control result includes meshing position difference, vibration error, and gear temperature difference; and the reducer optimization control result specifically includes corrected center of gravity value, rotational distribution coefficient, and updated window parameters.

[0013] As a further aspect of the present invention, the dynamic window module specifically comprises:

[0014] The data difference submodule collects the raw data of the reducer torque and speed, calculates the torque difference and speed difference at adjacent time points and summarizes them into the same time series, and performs statistical calculations based on the time series at a fixed sampling interval to generate the difference sequence interval;

[0015] The state estimation submodule calls the difference sequence interval, constructs a state division interval based on the torque difference and speed difference, uses a Markov chain model to calculate the transition probability of the numerical transition in adjacent intervals, analyzes the state change trend, and generates the state transition probability.

[0016] The window generation submodule, based on the state transition probability, combines the set transition time window and error range to perform numerical filtering of the matching interval, calculates the dynamic transition time length, and generates a dynamic transition window.

[0017] As a further aspect of the present invention, the temperature cooling module is specifically:

[0018] The temperature matching submodule acquires the reducer temperature data based on the dynamic transition window, divides the temperature data into window intervals and performs interval matching, calculates and records the mean value of the temperature points in each interval, and generates an interval average temperature sequence.

[0019] The trend judgment submodule calls the interval average temperature sequence, calculates the temperature change rate based on the mean difference between adjacent intervals, compares the temperature change rate with the temperature safe operation threshold, and generates a temperature change trend.

[0020] The power correction submodule, based on the temperature change trend, increases the cooling power by a fixed adjustment step size when the temperature value exceeds the safe operating threshold, and generates a cooling correction result.

[0021] The temperature safety operating threshold is set based on the temperature fluctuation range of the reducer during long-term operation and the tolerance of the equipment materials.

[0022] As a further aspect of the present invention, the lubrication pressure module specifically comprises:

[0023] The pressure acquisition submodule acquires lubricating oil pressure data and the current speed of the reducer, records the lubricating oil pressure data and speed values ​​synchronously, matches the two types of data at the same time point, and generates a synchronous data sequence.

[0024] The joint comparison submodule calls the synchronous data sequence and combines it with the cooling correction result to perform a joint comparison of the three types of values, calculates the joint ratio between lubricating oil pressure, speed and cooling correction power, and judges it with the lubricating oil pressure threshold, records the comparison status, and generates a joint comparison result.

[0025] The supply adjustment submodule, based on the joint comparison results, increases the supply of the lubricating pump when the judgment result does not reach the lubricating oil pressure threshold, and adds a correction coefficient to the original supply value to generate a lubricating supply adjustment command.

[0026] The lubricating oil pressure threshold is set by the rated output range of the lubricating pump, the oil circuit resistance loss, and the required lubricating film thickness on the gear meshing surface.

[0027] As a further aspect of the present invention, the gear meshing module specifically comprises:

[0028] The data acquisition submodule, based on the lubrication supply adjustment command, acquires the adjusted gear vibration data and gear temperature data, calculates the error for the vibration data and temperature data respectively, matches the error results with the time series and records them to generate an error value sequence.

[0029] The probability determination submodule calls the error value sequence, inputs the error data into the probability density function, analyzes the probability distribution coefficients of multiple intervals, calculates the centroid position based on the probability distribution coefficients and determines the degree of centroid offset, and generates the offset centroid.

[0030] The meshing adjustment submodule determines the difference between the offset center of gravity and the gear meshing reference position. When the offset difference exceeds the gear meshing reference, it adjusts the gear clearance in the reducer and generates the reducer control result.

[0031] The gear meshing reference is set by the gear structure parameters of the reducer, assembly process requirements, and operating status data.

[0032] As a further aspect of the present invention, in the process of calculating the centroid position based on the probability distribution coefficients, the multi-interval probability distribution coefficients in the interval probability distribution table are weighted and calculated with the median of the corresponding error value interval to obtain the weighted probability centroid position.

[0033] As a further aspect of the present invention, the optimization control module specifically comprises:

[0034] The rotation data submodule, based on the reducer control results, acquires the adjusted short-time sampled torque and speed data and performs distribution statistics, calculates the frequency ratio within the statistical interval, analyzes the joint distribution relationship between torque and speed, and obtains the rotation distribution coefficient.

[0035] The center of gravity correction submodule calls the rotation distribution coefficient, compares the distribution differences of multiple intervals in the rotation distribution coefficient with the probability center of gravity position of the control result, calculates the offset of the probability center of gravity, and superimposes the offset with the probability center of gravity position in the control result to generate a corrected center of gravity value.

[0036] The window update submodule performs sliding iterative updates on the corrected center of gravity value in conjunction with the dynamic transition window, records the sequence of numerical changes within the window during the iteration process, and generates the optimized control result of the reducer.

[0037] As a further aspect of the present invention, the sliding iterative update process specifically involves continuously iterating the corrected centroid value in the dynamic transition window according to a fixed sliding step size, and calculating the difference between the corrected centroid value of each iteration and the corrected centroid value at the previous time point during the iteration process.

[0038] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0039] In this invention, by collecting torque and speed data of the reducer and calculating the difference between adjacent time points, and combining this with a Markov chain model for state transition probability estimation, a transition time window and error range are set to effectively generate a dynamic transition window and transmit it to the subsequent control module, ensuring that the control process adapts to the real-time changes of the reducer. This approach improves the real-time performance and adaptability of the control system by dynamically adjusting control parameters and providing real-time feedback on reducer state changes. Simultaneously, temperature data is used for interval matching and trend judgment, effectively avoiding malfunctions caused by temperatures exceeding safe ranges and improving the stability of the reducer. A combined comparison of lubricating oil pressure and cooling correction results ensures suitable lubrication conditions, thereby avoiding wear and malfunctions due to insufficient lubrication. This refined control logic improves the system's accuracy and adaptability, avoiding operational errors that may result from single threshold control. By analyzing gear vibration and temperature data using probability density functions and adjusting gear meshing based on offset judgment, the entire reducer can maintain good operating efficiency under different loads and operating conditions, reducing the response lag of traditional control systems during load fluctuations and improving the robustness and efficiency of the system under complex operating conditions. Attached Figure Description

[0040] Figure 1 This is a system flowchart of the present invention;

[0041] Figure 2 This is a flowchart of the dynamic window module of the present invention;

[0042] Figure 3 This is a flowchart of the temperature cooling module of the present invention;

[0043] Figure 4 This is a flowchart of the lubrication pressure module of the present invention;

[0044] Figure 5 This is a flowchart of the gear meshing module of the present invention;

[0045] Figure 6 The flowchart for the optimized control module of this invention is shown below. Detailed Implementation

[0046] 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.

[0047] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0048] Please see Figure 1 A speed reducer control system based on chance constraints includes:

[0049] The dynamic window module collects the torque and speed data of the reducer and calculates the difference between adjacent time points. It estimates the state transition probability through a Markov chain model, sets the transition time window and error range, generates a dynamic transition window, and transmits it to the temperature cooling module.

[0050] The temperature cooling module acquires the gearbox temperature data based on the dynamic transition window and performs interval matching, calculates the temperature change trend and determines whether the temperature safety operating threshold is exceeded. If it is exceeded, the cooling power is increased, the cooling correction result is generated and transmitted to the lubrication pressure module.

[0051] The lubrication pressure module acquires lubricating oil pressure data and the current reducer speed, and performs a joint comparison with the cooling correction results. When the joint comparison result does not reach the lubricating oil pressure threshold, the lubrication pump supply is increased, a lubrication supply adjustment command is generated, and it is transmitted to the gear meshing module.

[0052] The gear meshing module, based on the lubrication supply adjustment command, acquires the adjusted gear vibration data and gear temperature data and calculates the error respectively, inputs the probability density function to determine the probability centroid offset, adjusts the gear meshing state and gear clearance in the reducer, and generates the reducer control result;

[0053] The optimized control module, based on the reducer control results, acquires the adjusted short-time sampled torque and speed data for distribution statistics, corrects the probability centroid of the reducer control results by offset, updates the dynamic transition window, and generates the optimized control results for the reducer.

[0054] The dynamic transition window includes state transition probability, transition time interval, and error allowable range. Cooling correction results include cooling power, temperature threshold difference, and temperature change rate. Lubrication supply adjustment commands include pump start / stop status and pressure compensation. Reducer control results include meshing position difference, vibration error, and gear temperature difference. The reducer optimization control results specifically include corrected center of gravity value, rotational distribution coefficient, and updated window parameters.

[0055] Please see Figure 2 The dynamic window module specifically includes:

[0056] The data difference submodule collects the raw data of the reducer torque and speed, calculates the torque difference and speed difference at adjacent time points and summarizes them into the same time series, and performs statistical calculations based on the time series at a fixed sampling interval to generate the difference sequence interval;

[0057] Raw data on the torque and speed of the reducer are collected in real time using torque and speed sensors mounted on the reducer shaft. The sampling frequency is set to 10 Hz. The torque unit is Newton-meter (Nm), and the speed unit is revolutions per minute (rpm). In actual industrial conveyor belt systems, changes in reducer load cause fluctuations in torque and speed. For example, at time t = 0.0 seconds, the torque sensor records a value of T = 100 Nm, and the speed sensor records a value of ω = 1500 rpm. At t = 0.1 seconds, T = 102 Nm. Nm, ω=1498rpm, then calculate the torque difference between adjacent time points. The action is to extract the torque values ​​at consecutive time points and perform subtraction operations. For example, calculate the torque difference ΔT(0.1)=T(0.1)-T(0.0)2Nm at t=0.1 seconds, and at the same time calculate the speed difference Δω(0.1)=ω(0.1)-ω(0.0)=-2rpm. In summary, the same time series assigns the torque difference and speed difference corresponding to each time point to the time point, forming a sequence structure. For example, the time series entry is in At t = 0.1 seconds, ΔT = 2 Nm and Δω = -2 rpm were recorded. Statistical calculations were performed based on the time series at fixed sampling intervals. The fixed sampling interval consisted of 5 data points per group, i.e., a time window of 0.5 seconds. The statistical calculations included calculating the arithmetic mean and standard deviation of the torque difference within each group. The torque difference data groups [2 Nm, 1 Nm, -1 Nm, 0 Nm, 2 Nm] at time points t = 0.1, 0.2, 0.3, 0.4, and 0.5 seconds were selected, and the mean μ-T = 0.8 Nm and the standard deviation was 1 were calculated. Similarly, the mean μω and standard deviation σω of the speed difference are calculated. The range of the difference sequence is defined by adding or subtracting the standard deviation from the mean. For example, the torque difference range is set to [μT-σT, μT+σT] = [-0.4Nm, 2.0Nm]. The speed difference range is calculated similarly. The range setting is based on the stability of historical data. For example, under normal operation, the difference fluctuation range should cover 95% of the data points. If another set of data is set with mean μ = 1.0 and standard deviation σ = 0.5, then the range is [0.5, 1.5].

[0058] Table 1: Example of raw data for the speed reducer

[0059] Time (s) Torque (Nm) Rotational speed (rpm) 0.0 100 1500 0.1 102 1498 0.2 101 1501 0.3 99 1499

[0060] Time (s) Torque (Nm) Rotational speed (rpm) 0.4 103 1502

[0061] As shown in Table 1, this table lists the raw torque and speed data of the reducer at 5 time points, which are used for subsequent difference calculations.

[0062] The state estimation submodule calls the difference sequence interval, constructs the state division interval based on the torque difference and speed difference, uses the Markov chain model to calculate the transition probability of the numerical transition in adjacent intervals, analyzes the state change trend, and generates the state transition probability.

[0063] The generated difference sequence range is called, for example, the torque difference range is [-0.4Nm, 2.0Nm], and the speed difference range is [-3rpm, 1rpm]. Based on the torque difference and speed difference, a state division range is constructed. The action is to divide the difference range into discrete intervals, each interval representing a state. For example, the torque difference is divided into three intervals: state 1 is (-0.4, 0.5)Nm, state 2 is (0.5, 1.5)Nm, and state 3 is (1.5, 2.0)Nm. The speed difference is divided similarly. The reference difference data distribution is set, for example, using equal-width binning, with the width adjusted based on the standard deviation. The minimum value of the speed difference data is set to -3 rpm, and the maximum value is set to 1 rpm, dividing it into three intervals: state A is (-3, -1) rpm, state B is (-1, 1) rpm, and state C is (1, 3) rpm. However, since the maximum value is 1 rpm, state C is empty, and the adjusted interval is (-3, -1) and (-1, 1). The action is to extract continuous state transition records from the time series, such as the number of times from state i to state j, Δω. j ω represents the magnitude of the rotational speed change per unit time within the state interval j. i ω represents the average speed difference within state interval i. j The value represents the average speed difference within state interval j, where i and j represent the numbers of different state intervals. The parameter X in the formula... t =i indicates that the state at time t is interval i, and the subscripts i and j represent the numbers of the state interval divisions, which are discrete integers. For example, i = 1 indicates state interval 1, Δω i This represents the speed change per unit time within state interval i, measured in revolutions per minute per second (rpm / s). It is obtained by extracting all speed difference data points belonging to state interval i from the difference sequence generated by the data difference submodule, calculating the absolute average of these data points, and then dividing by a fixed sampling interval Δt = 0.1s. For example, if the speed difference data points corresponding to state interval i are -2 rpm and -1.5 rpm, then the absolute average is (|-2| + |-1.5|) / 2 = 1.75 rpm, and the amplitude Δω i =1.75 / 0.1 = 17.5 rpm / s, ω i ω represents the average speed difference within state interval i, in rpm. It is obtained by calculating the arithmetic mean of the speed differences within this interval, as shown in the data points above. i = (-2 + (-1.5)) / 2 = -1.75 rpm, and obtain the parameter Δω of state j in the same way.j and ω j For example, in the denominator This represents the combined magnitude of the amplitude and difference of state i, intended to normalize the probability. Parameter values ​​are assigned based on actual monitoring data. The reasonable range for the speed difference is based on industrial reducer standards, ranging from -10 rpm to 10 rpm, and the reasonable range for the amplitude is from 0 rpm / s to 100 rpm / s (the maximum difference is set to 10 rpm divided by Δt = 0.1 s). Parameter values ​​are obtained through example data. The speed difference data points corresponding to state interval i = 1 are from a time series, such as points -2 rpm, -1.5 rpm, and -1.8 rpm. Δω is calculated. i At that time, the absolute average value is (2 + 1.5 + 1.8) / 3 = 1.767 rpm, and the amplitude is 1.767 / 0.1 = 17.67 rpm / s, rounded to Δω. i =17.7 rpm / s, ω i =(-2-1.5-1.8) / 3 = -1.767 rpm, rounded to ω i = -1.77rpm, state interval j=2 corresponds to data points 3rpm, 4rpm, 5rpm, Δω j = (3+4+5) / 3 / 0.1 = 12 / 0.1 = 120 rpm / s, but the upper limit of the reasonable range is 100 rpm / s. Adjust the data points to 2 rpm, 3 rpm, and 4 rpm, then Δω j =(2+3+4) / 3 / 0.1=9 / 0.1=90rpm / s,ω j = (2+3+4) / 3 = 3 rpm, set parameter value Δω i =17.7 rpm / s, ω i = -1.77 rpm, Δω j =90rpm / s, ω j =3 rpm. Substitute into the formula to calculate: Molecules|Δω i -Δω j |+|ω i -ω j |=|17.7-90.0|+|-1.77-3.00|=72.3+4.77=77.07, denominator Therefore, Pr = 77.07 / 18.79 ≈ 4.10, but the probability value should not be greater than 1, indicating that the formula needs to be adjusted or the parameter settings need to be optimized. In practice, this can be adjusted by limiting the denominator or using the sigmoid function. The trend of state change is analyzed by comparing the probability values ​​of different state transitions. For example, a high probability indicates frequent transitions, and state transition probabilities are generated.

[0064] The window generation submodule, based on the state transition probability, combines the set transition time window and error range to perform numerical filtering of the matching interval, calculates the dynamic transition time length, and generates a dynamic transition window.

[0065] Based on the generated state transition probabilities, for example, the probability of transitioning from state i to j is 4.10, but it needs to be normalized. A transition time window and error range are used to filter for matching intervals. The initial transition time window is set to 2 seconds, and the error range is set to a probability deviation of 0.1. The action is to compare the transition probabilities with a threshold. The threshold is set with reference to the system response time; for example, a probability greater than 0.8 is considered a significant transition. The error range is based on historical fluctuations. The state transition probability list contains values ​​of 0.9, 0.5, and 0.7. The matching interval is selected for transitions with a probability greater than 0.8, i.e., selecting the state pair corresponding to 0.9. The dynamic transition time length is calculated by integrating the probability over the time window, for example, summing or averaging the probability values ​​within the window. Generating the dynamic transition window involves defining a time interval, such as the time period from the start to the end of the transition. In this example, the selected transition time points are set from t=1 second to t=3 seconds, so the dynamic transition window length is 2 seconds.

[0066] Please see Figure 3 The temperature cooling module is specifically as follows:

[0067] The temperature matching submodule acquires the reducer temperature data based on the dynamic transition window, divides the temperature data into window intervals and performs interval matching, calculates and records the mean value of the temperature points in each interval, and generates an interval average temperature sequence.

[0068] The reducer temperature data is acquired based on a dynamic transition window, such as a list of window start and end times. For example, window p=1 corresponds to the time interval [0 seconds, 2 seconds], and window p=2 corresponds to the time interval [2 seconds, 4 seconds]. These windows are generated from the dynamic transition window results mentioned earlier. The window data is stored in the form of a timestamp array, such as window 1: start=0s, end=2s, window 2: start=2s, end=4s. Simultaneously, reducer temperature data is acquired in real time from a temperature sensor. The sampling frequency is set to 1 Hz, i.e., one data point per second. The temperature unit is degrees Celsius (°C). The sensor type is PT100 RTD, with a measurement range of 0-150°C and an accuracy of ±0.5°C. In actual industrial gearbox cooling systems, temperature data reflects the thermal state of the equipment. For example, in a time series, the temperature is 60℃ at t=0 seconds, 61℃ at t=1 seconds, 62℃ at t=2 seconds, 63℃ at t=3 seconds, and 64℃ at t=4 seconds. The temperature data is divided into window intervals. The process involves reading the time boundary of each window, comparing the timestamp of the temperature data point with the window boundary, and assigning the temperature points whose timestamps fall within the window to the corresponding window. For example, for window p=1, with a time interval [0s, 2s], the temperature points include t=0s, 1s, and 2s. However, attention must be paid to window boundary handling. If the window is left-closed and right-open, then t=2s belongs to window p=2. (The last sentence appears to be incomplete and requires further context.) If the interval is closed, then window p=1 contains points t=0s, 1s, 2s, and window p=2 contains points t=2s, 3s, 4s. However, t=2s may overlap, so a unique assignment needs to be defined. Typically, a left-closed, right-open approach is used: window p=1: (0s, 2s) contains t=0s, 1s, and window p=2: (2s, 4s) contains t=2s, 3s. Interval matching is performed to ensure that each temperature point is assigned to only one window, avoiding duplication. This is done by comparing the timestamp and window boundary values, using less than or greater than operators. For example, for a timestamp t, if start ≤ t < end, then it is assigned to a window. In the example, t=0s and t=1s are assigned to window p=1, and t=2s and t=3s are assigned to window p=2. The temperature value at t=3s is assigned to window p=2, and t=4s may belong to a subsequent window. The mean value of the temperature points within each interval is calculated. The action is to extract all temperature values ​​assigned to the window, calculate the arithmetic mean of these values, that is, sum them and divide by the number of points. For example, if window p=1 has temperature points 60℃ and 61℃, the mean value is (60+61) / 2 = 60.5℃. If window p=2 has temperature points 62℃ and 63℃, the mean value is (62+63) / 2 = 62.5℃. These values ​​are recorded. The action is to store the mean value of each window into an array or list, arranged in window order. T-mean = [60.5℃, 62.5℃, 64.5℃] corresponds to windows p=1, 2, 3, generating an interval mean temperature sequence.

[0069] The trend judgment submodule calls the interval average temperature sequence, calculates the temperature change rate based on the difference in mean between adjacent intervals, compares the temperature change rate with the temperature safe operation threshold, and generates a temperature change trend.

[0070] The generated interval mean temperature sequence is called, for example, the sequence is T-mean = [60.5℃, 62.5℃, 64.5℃] with corresponding windows p = 1, 2, 3. The rate of temperature change is calculated based on the difference in mean between adjacent intervals. The mean of consecutive windows in the sequence is extracted, and the rate of change between them is calculated, but the formula uses a weighted method. In the diagram, parameter Rp represents the rate of temperature change between adjacent intervals p and p+1, measured in °C per second, and parameter T... p，q This represents the average temperature at time q within the p-th interval, in °C. The subscript p indicates the window number, which is a discrete integer, such as p = 1, 2, 3. The subscript q indicates the time number, from 1 to n. The parameter T... p+1，q The parameter W represents the average temperature at time q within the (p+1)th interval. q The parameter q represents the temperature weighting coefficient at time q, which is dimensionless. The parameter n represents the total number of temperature sampling times within the interval, and is an integer. The parameter values ​​are based on actual data, and the reasonable temperature range refers to industrial standards. The reducer's operating temperature is 30-100℃. The total number of sampling times n is calculated using the window duration and sampling interval. With a window duration of 2 seconds and a sampling interval of 1 second, n = 2. p，q The values ​​are obtained from a temperature sensor. For example, window p=1 corresponds to the time interval (0s, 2s), time q=1 is t=0s temperature=60℃, q=2 is t=1s temperature=61℃, window p=2 corresponds to (2s, 4s), q=1 is t=2s temperature=62℃, q=2 is t=3s temperature=63℃, and the weight W... q Set the importance of reference times, such as linear weights, so that later times have higher weights. The quantization standard is W. q =q / n, but needs to be normalized. In the example, n=2, W1=0.5, W2=1.0. The weighting coefficients are set based on experience, with recent temperature changes having a greater impact, so the weights increase incrementally. Through the example, the parameter value n=2, T 1，1 =60℃, T 1，2 =61℃, T 2，1 =62℃, T 2，2 =63℃, W1=0.5, W2=1.0, calculate Rp for p=1, molecule The absolute value |-3| = 3, and the denominator is... Therefore, R1 = 3 / 1.118 ≈ 2.683℃ / s. The advantage of this formula lies in the introduction of a weighting coefficient W.q Differentiated processing of temperature values ​​at different times improves the accuracy of the rate of change calculation. The results show that the rate value of 2.683℃ / s is lower than the safety threshold of 5℃ / s, which means that the temperature change is within the safe range. The numerical results are directly used for trend judgment, and a "stable" trend is generated by comparison.

[0071] Table 2: Example Table of Temperature Data and Weights

[0072] Window p Time q Temperature T{p, q} Weight Wq 1 1 60 0.5 1 2 61 1.0

[0073] Window p Time q Temperature T{p, q} Weight Wq 2 1 62 0.5 2 2 63 1.0

[0074] As shown in Table 2, this table lists the temperature values ​​and weighting coefficients for the two windows used to calculate the rate of temperature change.

[0075] The power correction submodule, based on the temperature change trend, increases the cooling power by a fixed adjustment step size when the temperature value exceeds the safe operating threshold, and generates a cooling correction result.

[0076] The safe operating temperature threshold is set based on the temperature fluctuation range of the reducer during long-term operation and the tolerance of the equipment materials.

[0077] Based on the generated temperature change trend, such as a "stable" trend or a numerical indicator, when the temperature value exceeds the safe operating threshold, the temperature value refers to the average value in the interval temperature series or the real-time temperature. The safety threshold is set with reference to the equipment's rated value, such as the maximum safe temperature of the reducer being 80℃. Based on the insulation class, the threshold is obtained through experimental testing. For example, in a load test, a temperature stable below 75℃ is considered safe, therefore the threshold = 75℃. The action is to compare the temperature value with the threshold using the greater than operator. For example, the latest average value in the interval temperature series is 64.5℃, compared with the threshold of 75℃. 64.5 < 75, not exceeded, but the trend may indicate future risks. The cooling power is increased according to a fixed adjustment step size. The fixed adjustment step size is set with reference to the cooling system capacity, such as a step size = 10% of the rated power, set according to the manufacturer's recommendations. In this example, the rated cooling power = 1000W, the step size = 100W, and the increase action is to increase the power output value. For example, the current power = 500W, after the increase = 600W, generating a cooling correction result.

[0078] Please see Figure 4 The lubrication pressure module is specifically as follows:

[0079] The pressure acquisition submodule acquires lubricating oil pressure data and the current speed of the reducer, records the lubricating oil pressure data and speed values ​​synchronously, matches the two types of data at the same time point, and generates a synchronous data sequence.

[0080] The system acquires lubricating oil pressure data and the current reducer speed. Pressure data is collected in real-time by a pressure sensor installed in the lubricating oil circuit, measured in bar (0-20 bar) with an accuracy of ±0.1 bar. Simultaneously, reducer speed data is collected via a shaft-end encoder, measured in revolutions per minute (rpm), with a measurement range of 0-3000 rpm and an accuracy of ±1 rpm. The sampling frequency is set to 10 Hz, i.e., the time interval Δt = 0.1 seconds. In actual industrial reducer lubrication systems, pressure data reflects the oil circuit status, and speed data reflects... The equipment's operating speed, for example, at time point t = 0.0 seconds, the pressure sensor records a value of P = 10.0 bar, and the speed sensor records a value of N = 1500 rpm; at t = 0.1 seconds, P = 10.1 bar and N = 1498 rpm; at t = 0.2 seconds, P = 10.2 bar and N = 1501 rpm. The lubricating oil pressure data and speed values ​​are recorded synchronously. The execution action uses the system clock to assign a timestamp to each data point to ensure time consistency. For example, the pressure data point is accompanied by a timestamp t = 0.0. The data point is set to s, with a value of 10.0 bar and a speed data point with a timestamp of t = 0.0 s and a value of 1500 rpm. This performs a matching operation on two types of data at the same time point, comparing the timestamps for equality. The equals operator is used to pair pressure values ​​and speed values ​​with the same timestamp. For example, at t = 0.0 s, the pressure of 10.0 bar and the speed of 1500 rpm are matched, forming the data pair (10.0, 1500). At t = 0.1 s, the pressure of 10.1 bar and the speed of 1498 rpm are matched, forming the data pair (...). In the data acquisition process (10.1, 1498), the pressure value is obtained by converting the sensor's analog signal through an ADC, and noise is removed by digital filtering. The rotational speed value is calculated by encoder pulse counting. For example, when t = 0.3s, P = 10.3bar and N = 1502rpm, the matching is (10.3, 1502). At time point t = 0.0s, the pressure P = 10.0bar and the rotational speed N = 1500rpm. At t = 0.1s, P = 10.1bar and N = 1498rpm, generating a synchronous data sequence.

[0081] Table 3: Example Table of Synchronized Data

[0082] Time (s) Lubricating oil pressure (bar) Rotational speed (rpm) 0.0 10.0 1500 0.1 10.1 1498 0.2 10.2 1501 0.3 10.3 1502

[0083] As shown in Table 3, this table lists the synchronous data of lubricating oil pressure and speed at four time points for subsequent comparative analysis.

[0084] The joint comparison submodule calls the synchronous data sequence and combines it with the cooling correction result to perform a joint comparison of the three types of values, calculates the joint ratio between lubricating oil pressure, speed and cooling correction power, judges it against the lubricating oil pressure threshold, records the comparison status, and generates a joint comparison result.

[0085] The system retrieves synchronized data sequences and combines them with cooling correction results. For example, the sequence might contain values ​​at time t = 0.0s (pressure P = 10.0 bar, speed N = 1500 rpm) and at t = 0.1s (P = 10.1 bar, N = 1498 rpm). Simultaneously, it acquires cooling correction results, such as cooling power W = 600W. Based on these corrections, if the power increases from 500W to 600W, the three values ​​are compared together. The action involves extracting the pressure and speed values ​​from the synchronized data sequence and combining them with the cooling power value. For instance, at time t = 0.0s, the three values ​​are...

[0086] Given P = 10.0 bar, N = 1500 rpm, and W = 600 W, calculate the combined ratio between lubricating oil pressure, rotational speed, and cooling power correction. The defined formula is ratio = P / (N*W). However, the user requested that the unmentioned formula not be included, so the calculation process is described in words: pressure value divided by the product of rotational speed and power values. For example, at t = 0.0 s, the product N*W = 1500*600 = 900000, and the ratio is 0.00001111 bar / (rpm·W). The units are simplified; in practice, normalization may be necessary. The ratio is then compared with the lubricating oil pressure threshold. The lubricating oil pressure threshold is set according to the lubrication system requirements, such as a minimum pressure threshold P-threshold of 5 bar. Based on the oil pump performance, the threshold is obtained from the manufacturer's documentation. In this example, the threshold is 5 bar. The judgment action is to compare the pressure value P with the threshold P-threshold using the greater than or equal to operator. For example, at t = 0.0 s, P = 10.0 bar > 5 bar, therefore it meets the standard. The comparison status is recorded. The action to be executed is to set a status flag based on the judgment result, such as "meets the standard" or "does not meet the standard". For example, at t = 0.0 s, the status is "meets the standard", and a joint comparison result is generated.

[0087] The supply adjustment submodule increases the lubrication pump supply when the judgment result does not reach the lubrication oil pressure threshold, based on the joint comparison result. It adds a correction coefficient to the original supply value and generates a lubrication supply adjustment command.

[0088] The lubricating oil pressure threshold is set by the rated output range of the lubrication pump, the oil circuit resistance loss, and the required lubricating film thickness on the gear meshing surface.

[0089] Based on the joint comparison results, such as the latest status in the status sequence being "meets the standard" or "does not meet the standard," when the judgment result does not reach the lubricating oil pressure threshold (meaning the pressure value is less than the threshold, e.g., P < P - threshold), where the lubricating oil pressure threshold P - threshold = 5 bar), the action is to compare the pressure value with the threshold using the less than operator. For example, if at a certain time point P = 4.5 bar < 5 bar, an adjustment is triggered, increasing the lubricating pump supply. The lubricating pump supply unit is liters per minute (L / min). The original supply value is set based on the system default, such as the original supply value S. -original = 10L / min. Based on the pump specifications, a correction factor is added to the original supply value. The correction factor is set with reference to the adjustment range. For example, the correction factor C = 0.5L / min. Based on empirical values, the factor is obtained through calibration tests. For example, in the test, increasing the supply by 0.5L / min can increase the pressure by 0.1 bar. The accumulation action is to calculate the new supply S-new = S-original + C. For example, if S-original = 10L / min and C = 0.5L / min, then S-new = 10.5L / min, and a lubrication supply adjustment command is generated.

[0090] Please see Figure 5 The gear meshing module is specifically as follows:

[0091] The data acquisition submodule acquires gear vibration data and gear temperature data after adjustment based on the lubrication supply adjustment command, calculates the error for the vibration data and temperature data respectively, matches the error results with the time series and records them to generate an error value sequence.

[0092] Based on lubrication supply adjustment commands, such as setting the lubrication pump supply to 10.5 L / min, and based on the previous lubrication supply adjustment command result, such as increasing the supply from 10 L / min to 10.5 L / min, the adjusted gear vibration and gear temperature data are acquired. The action is to call the vibration acceleration sensor installed in the gearbox to collect vibration data in real time, in meters per second squared (m / s²). 2 Measurement range: 0-20 m / s 2 Accuracy ±0.01m / s 2 The sampling frequency is 100 Hz, and a gear temperature sensor is used to collect temperature data in degrees Celsius (°C). The measurement range is 0-150°C, with an accuracy of ±0.5°C and a sampling frequency of 1 Hz. In actual industrial reducer monitoring systems, vibration data reflects the gear meshing state, and temperature data reflects the thermal load. For example, at time t = 1.0 seconds after adjusting the supply, the vibration sensor records a value of V = 0.55 m / s. 2 The temperature sensor recorded a value of T = 61℃, and at t = 1.1 seconds, V = 0.56 m / s.2 T = 61.2℃. Errors are calculated for both vibration and temperature data. The action to be executed is to set a reference value or desired value. The vibration reference value V-ref = 0.5 m / s². 2 Based on an ideal stable operating condition, the vibration average is obtained through historical data averaging, for example, the average vibration value calculated from no-load tests is 0.5 m / s. 2 The temperature reference value T-ref = 60℃ is based on rated operating conditions and obtained from the manufacturer's specifications. Calculation errors are performed using subtraction. For example, at t = 1.0 seconds, EV = 0.55 - 0.5 = 0.05 m / s. 2 ET = 61 - 60 = 1℃, and at t = 1.1 seconds, the vibration error EV = 0.56 - 0.5 = 0.06 m / s. 2 The temperature error ET = 61.2 - 60 = 1.2℃. The error result is matched with the time series and recorded. The action is to assign a timestamp to each error value, consistent with the data acquisition time, using the equals operator to match the same time points. For example, at t = 1.0 seconds, a vibration error of 0.05 m / s is matched. 2 A temperature error of 1℃ forms an error pair (0.05, 1.0). At t = 1.1 seconds, the pair matches (0.06, 1.2). The recording process involves storing these error pairs in a list or array in chronological order. In the data acquisition process, vibration values ​​are obtained from the sensor's analog signal via ADC conversion and digital filtering smoothing. Temperature values ​​are read directly. The reference value setting process involves calibration; for example, in standard testing, the vibration value is stabilized at 0.5 m / s². 2 Therefore, let V-ref = 0.5. In the example calculation, when t = 1.2 seconds, V = 0.57 m / s. 2 If T = 61.5℃, then EV = 0.07 m / s 2 With ET = 1.5℃ and matching records, the vibration error EV = 0.05 m / s at time point t = 1.0 seconds. 2 With a temperature error ET = 1.0℃ and EV = 0.06m / s at t = 1.1 seconds,... 2 ET = 1.2℃, generating an error numerical sequence.

[0093] The probability determination submodule calls the error value sequence, inputs the error data into the probability density function, analyzes the probability distribution coefficients of multiple intervals, calculates the centroid position based on the probability distribution coefficients, determines the degree of centroid offset, and generates the offset centroid.

[0094] Call the error value sequence, for example, the sequence contains the vibration error EV = 0.05 m / s at time point t = 1.0 seconds. 2 With a temperature error ET = 1.0℃ and EV = 0.06m / s at t = 1.1 seconds,... 2ET = 1.2℃. The error data is input into the probability density function. The action defines the data range of the error value, such as a vibration error range of -0.1 to 0.1 m / s². 2 The temperature error range is -5 to 5℃, and this range is divided into multiple intervals. For example, for vibration error, the intervals are (-0.1, -0.05), (-0.05, 0), (0, 0.05), and (0.05, 0.1), with an interval width of 0.05 m / s. 2 For each interval, calculate the probability that the error value falls within that interval. The probability is calculated as the number of data points within the interval divided by the total number of data points. For example, with 3 total data points, vibration error values ​​of 0.05, 0.06, and 0.07 all fall within the interval (0.05, 0.1), so the probability for that interval is 3 / 3 = 1.0. However, the probability should be less than 1. Adjust the data points by setting the sequence to have more points, such as adding a point at t = 1.3 seconds and EV = 0.04 m / s. 2 If the value falls within (0, 0.05), there are a total of 4 points, with 3 points in the interval (0.05, 0.1). The probability is 3 / 4 = 0.75. The probability distribution coefficients of multiple intervals are analyzed. The probability distribution coefficients represent the probability value of each interval. The action is to record the probability of each interval. For example, the probability of the vibration error interval is: (-0.1, -0.05) = 0, (-0.05, 0) = 0, (0, 0.05) = 0.25, (0.05, 0.1) = 0.75. Based on the probability distribution coefficients, the center of gravity position is calculated using the weighted average formula: Center of gravity G = ∑(interval midpoint * probability) / ∑ probability. Calculate the midpoint value of each interval. For example, in the interval (0, 0.05), the midpoint is 0.025, the probability is 0.25, and the product is 0.025 * 0.25 = 0.00625. Calculate other intervals similarly. After summing, divide by the sum of probabilities, which should equal 1. For example, in calculating the centroid of vibration error, the midpoints of the intervals are: (-0.1, -0.05) midpoint = -0.075, prob = 0; (-0.05, 0) midpoint = -0.025, prob = 0; (0, 0.05) midpoint = 0.025, prob = 0.25; (0.05, 0.1) midpoint = 0.075, prob = 0.75.

[0095] Weighted sum = (-0.0750) + (-0.0250) + (0.025 * 0.25) + (0.075 * 0.75) = 0.0625, probability sum = 0.25 + 0.75 = 1.0, centroid G = 0.0625 / 1.0 = 0.0625 m / s 2 The degree of center of gravity shift is determined. The degree of shift refers to the absolute difference between the center of gravity and the zero point, i.e., shift = |G-0|. For example, if the vibration error center of gravity is 0.0625, the shift = 0.0625. A shift threshold is set, such as threshold = 0.02m / s². 2Based on stability requirements, the threshold is obtained experimentally. For example, in the test, an offset less than 0.02 indicates stability. The judgment action is to compare the offset with the threshold using the greater than operator. For example, if the offset is 0.0625 > the threshold is 0.02, the offset is significant, and the offset centroid is generated. The result is the centroid value and the offset flag. For example, the offset centroid is (0.0625, significant). In data acquisition, the error sequence is from the previous text. The interval division is based on the data range, and the probability calculation is done by counting. Temperature error is handled similarly. The temperature error values ​​are set to 1.0, 1.2, and 1.5, with a range of -5 to 5℃. The intervals are divided into (0, 2.5) and [2.5, 5]. The midpoints are 1.25 and 3.75. The probability is calculated. All points fall within (0, 2.5), with a probability of 1.0, a centroid of 1.25, and an offset of 1.25. The threshold is set to 1.0℃. Comparing 1.25 > 1.0, the offset is significant, and the offset centroid is generated.

[0096] The meshing adjustment submodule determines the difference between the offset center of gravity and the gear meshing reference position. When the offset difference exceeds the gear meshing reference, it adjusts the gear clearance in the reducer and generates the reducer control result.

[0097] The gear meshing reference is set by the gear structure parameters of the reducer, assembly process requirements, and operating status data.

[0098] Based on the offset of the centroid, for example, the vibration error offset of the centroid is (0.0625m / s). 2 The temperature error offset from the center of gravity is (1.25℃, significant). This is used to determine the difference from the gear meshing reference position. The gear meshing reference position is set to zero, meaning the ideal error value is 0. The difference refers to the value of the offset from the center of gravity, such as a vibration offset of 0.0625 m / s. 2 The temperature deviation is 1.25℃. When the deviation difference exceeds the gear meshing reference, "exceeds" means the deviation value is greater than the reference tolerance. The reference tolerance is set based on the allowable gear clearance value, such as vibration tolerance = 0.02m / s. 2 The temperature tolerance is 1.0℃, obtained from the manufacturer's specifications. The action involves comparing the offset value with the tolerance using the greater than operator. For example, vibration offset 0.0625 > tolerance 0.02, temperature offset 1.25 > tolerance 1.0, therefore both exceed the tolerance. The gear clearance within the reducer is then adjusted. The adjustment action modifies the gear clearance setting value in millimeters (mm). The original clearance value is set as the base value, such as clearance - base = 0.1mm, obtained from the design value. The adjustment amount is based on the degree of offset using a correction factor, such as correction factor K = 0.01, obtained through calibration testing. For example, for every 0.01m / s increase in offset... 2 Adjust the gap by 0.01m, and calculate the new gap: gap base + K * offset. For example, for a vibration offset of 0.0625m / s... 2The new clearance = 0.1 + 0.01 * 0.0625 = 0.1 + 0.000625 = 0.100625 mm. Similar temperature deviations may affect thermal expansion, but are mainly based on vibration, generating the reducer control results.

[0099] Please see Figure 6 The optimized control module specifically includes:

[0100] The rotation data submodule, based on the reducer control results, acquires the adjusted short-time sampled torque and speed data and performs distribution statistics, calculates the frequency ratio within the statistical interval, analyzes the joint distribution relationship between torque and speed, and obtains the rotation distribution coefficient.

[0101] Based on the reducer control results, for example, if the control result is to set the gear backlash to 0.100625 mm, the system acquires short-sampling torque and speed data after adjustment. The action executed is to call the torque sensor to collect the torque value (unit: N·m, measurement range: 0-1000 N·m, accuracy: ±1 N·m, sampling frequency: 10 Hz), and simultaneously call the speed sensor to collect the speed value (unit: revolutions per minute, measurement range: 0-3000 revolutions per minute, accuracy: ±1 revolution per minute), with the short-sampling window set to 2 seconds. There are a total of 20 data points. In actual industrial gearbox monitoring, for example, at time point t = 0.0 seconds, the recorded torque value is 500 N·m and the recorded speed value is 1500 rpm. Sampling continues until t = 2.0 seconds, a data point list is obtained, and distribution statistics are performed. The action executed is to define statistical intervals: the torque interval is divided into 0-200, 200-400, 400-600, 600-800, and 800-1000 N·m, with an interval width of 200 N·m; the speed interval is divided into 0-10 N·m. The engine speed ranges from 00 to 3000 RPM, with intervals of 1000 RPM. The frequency of data points within each interval is calculated. Greater than / equal to and less than / less than operators are used to determine if a value falls within an interval. For example, if a torque value of 500 Nm falls within the 400-600 Nm interval, the frequency is incremented by 1. If 12 out of 20 torque points fall within the 400-600 Nm interval, the frequency is 12. The proportion of values ​​within each interval is calculated, and the action is performed for each interval, with the proportion set to... Interval frequency / total number of points, with a total of 20 points. For example, the ratio of torque interval 400-600 N·m = 12 / 20 = 0.6. Analyze the joint distribution relationship between torque and speed. The action is to define a joint interval, such as the combination of torque 400-600 N·m and speed 1000-2000 rpm. Count the frequency of data points that fall into both intervals. For example, if 10 out of 20 points meet the criteria, the joint frequency = 10. Calculate the joint frequency ratio = 10 / 20 = 0.5 to obtain the rotational distribution coefficient.

[0102] The center of gravity correction submodule calls the rotation distribution coefficient, compares the distribution differences of multiple intervals in the rotation distribution coefficient with the probabilistic center of gravity position of the control result, calculates the offset of the probabilistic center of gravity, and superimposes the offset with the probabilistic center of gravity position in the control result to generate the corrected center of gravity value.

[0103] Based on the rotational distribution coefficient, for example, a coefficient value of 0.5, the rotational distribution coefficient is invoked. The action performed is to read the coefficient value and, based on the probability centroid position of the control result (obtained from the probability determination submodule mentioned earlier, for example, the vibration error probability centroid is 0.0625 m / s², and the temperature error probability centroid is 1.25 degrees Celsius), compare the distribution differences across multiple intervals in the rotational distribution coefficient. The action performed is to define the coefficient value range of 0-1 into intervals such as 0-0.2, 0.2-0.4, 0.4-0.6, 0.6-0.8, and 0.8-1.0, with an interval width of 0.2, and calculate the absolute difference between the coefficient value and the midpoint of the interval. For example, the coefficient value is calculated based on the probability centroid position of the vibration error. The value 0.5 falls within the range of 0.4-0.6, with a midpoint of 0.5 and a difference of 0. The offset of the probability centroid is calculated, where the offset is the difference between the probability centroid and the desired centroid. The desired centroid is set to zero. For example, if the desired centroid for vibration error is 0 m / s², the offset is 0.0625 m / s². The offset is then superimposed on the probability centroid position in the control results for correction. The action is to set the correction coefficient K = 0.1, which is set empirically within a reasonable range of 0-1. The correction value = probability centroid + K offset. For example, the vibration error correction value = 0.0625 + 0.1 * 0.0625 = 0.06875 m / s², generating the corrected centroid value.

[0104] The window update submodule, for the corrected center of gravity value, combines a dynamic transition window to perform sliding iterative updates on the corrected center of gravity value, records the sequence of numerical changes in the window during the iteration process, and generates the optimized control results of the reducer.

[0105] Based on the corrected center of gravity value, for example, a vibration error corrected center of gravity value of 0.06875 meters per square second, the action performed is to read this value and then iteratively update the corrected center of gravity value using a dynamic transition window. The dynamic transition window is obtained from the window generation submodule mentioned earlier. For example, the window parameters are a list of time intervals, with window 1 corresponding to (0 seconds, 2 seconds), window 2 corresponding to (2 seconds, 4 seconds), and the window length being 2 seconds. The sliding iterative update action involves distributing the corrected center of gravity value to the windows in chronological order, calculating the average value of the center of gravity value within each window, and then sliding to the next window. For example, at time t = 2 seconds, window 1 contains a sequence of corrected centroid values, with set values ​​of 0.06875, 0.06900, 0.06850 meters per square second, etc. The mean within the window is calculated, such as a mean of 0.06875 meters per square second. Then, starting from t = 2 seconds, window 2 incorporates the new values, recalculates the mean, and records the sequence of numerical changes within the window during the iteration process. The action performed is to store the mean of each window into the sequence, for example, the sequence is (window 1 mean = 0.06875, window 2 mean = 0.06900, etc.), generating the optimized control result of the reducer.

[0106] 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. A speed reducer control system based on chance constraints, characterized in that, The system includes: The dynamic window module collects the torque and speed data of the reducer and calculates the difference between adjacent time points. It estimates the state transition probability through the Markov chain model, sets the transition time window and error range, generates a dynamic transition window, and transmits it to the temperature cooling module. The dynamic window module is specifically: The data difference submodule collects the raw data of the reducer torque and speed, calculates the torque difference and speed difference at adjacent time points and summarizes them into the same time series, and performs statistical calculations based on the time series at a fixed sampling interval to generate the difference sequence interval; The state estimation submodule calls the difference sequence interval, constructs a state division interval based on the torque difference and speed difference, uses a Markov chain model to calculate the transition probability of the numerical transition in adjacent intervals, analyzes the state change trend, and generates the state transition probability. The window generation submodule, based on the state transition probability, combines the set transition time window and error range to perform numerical filtering of the matching interval, calculates the dynamic transition time length, and generates a dynamic transition window. The temperature cooling module acquires the reducer temperature data based on the dynamic transition window and performs interval matching, calculates the temperature change trend and determines whether the temperature safety operating threshold is exceeded. If it is exceeded, the cooling power is increased, a cooling correction result is generated and transmitted to the lubrication pressure module. The lubrication pressure module acquires lubricating oil pressure data and the current speed of the reducer, and performs a joint comparison with the cooling correction result. When the joint comparison result does not reach the lubricating oil pressure threshold, the lubrication pump supply is increased, a lubrication supply adjustment command is generated, and the command is transmitted to the gear meshing module. The gear meshing module, based on the lubrication supply adjustment command, acquires the adjusted gear vibration data and gear temperature data and calculates the error respectively, inputs the probability density function to determine the probability centroid offset, adjusts the gear meshing state and gear clearance in the reducer, and generates the reducer control result; The gear meshing module is specifically: The data acquisition submodule, based on the lubrication supply adjustment command, acquires the adjusted gear vibration data and gear temperature data, calculates the error for the vibration data and temperature data respectively, matches the error results with the time series and records them to generate an error value sequence. The probability determination submodule calls the error value sequence, inputs the error data into the probability density function, analyzes the probability distribution coefficients of multiple intervals, calculates the centroid position based on the probability distribution coefficients and determines the degree of centroid offset, and generates the offset centroid. The meshing adjustment submodule determines the difference between the offset center of gravity and the gear meshing reference position. When the offset difference exceeds the gear meshing reference, it adjusts the gear clearance in the reducer and generates the reducer control result. The gear meshing reference is set by the gear structure parameters of the reducer, assembly process requirements, and operating status data. The dynamic transition window includes state transition probability, transition time interval, and error allowable range; the cooling correction result includes cooling power, temperature threshold difference, and temperature change rate; the lubrication supply adjustment command includes pump start / stop status and pressure compensation amount; and the reducer control result includes meshing position difference, vibration error, and gear temperature difference.

2. The speed reducer control system based on chance constraints according to claim 1, characterized in that, The temperature cooling module is specifically: The temperature matching submodule acquires the reducer temperature data based on the dynamic transition window, divides the temperature data into window intervals and performs interval matching, calculates and records the mean value of the temperature points in each interval, and generates an interval average temperature sequence. The trend judgment submodule calls the interval average temperature sequence, calculates the temperature change rate based on the mean difference between adjacent intervals, compares the temperature change rate with the temperature safe operation threshold, and generates a temperature change trend. The power correction submodule, based on the temperature change trend, increases the cooling power by a fixed adjustment step size when the temperature value exceeds the safe operating threshold, and generates a cooling correction result. The temperature safety operating threshold is set based on the temperature fluctuation range of the reducer during long-term operation and the tolerance of the equipment materials.

3. The speed reducer control system based on chance constraints according to claim 1, characterized in that, The lubrication pressure module is specifically: The pressure acquisition submodule acquires lubricating oil pressure data and the current speed of the reducer, records the lubricating oil pressure data and speed values ​​synchronously, matches the two types of data at the same time point, and generates a synchronous data sequence. The joint comparison submodule calls the synchronous data sequence and combines it with the cooling correction result to perform a joint comparison of the three types of values, calculates the joint ratio between lubricating oil pressure, speed and cooling correction power, and judges it with the lubricating oil pressure threshold, records the comparison status, and generates a joint comparison result. The supply adjustment submodule, based on the joint comparison results, increases the supply of the lubricating pump when the judgment result does not reach the lubricating oil pressure threshold, and adds a correction coefficient to the original supply value to generate a lubricating supply adjustment command. The lubricating oil pressure threshold is set by the rated output range of the lubricating pump, the oil circuit resistance loss, and the required lubricating film thickness on the gear meshing surface.

4. The speed reducer control system based on chance constraints according to claim 1, characterized in that, In the process of calculating the centroid position based on the probability distribution coefficients, the multi-interval probability distribution coefficients in the interval probability distribution table are weighted and calculated with the median of the corresponding error value interval to obtain the weighted probability centroid position.

5. A speed reducer control system based on chance constraints according to claim 1, characterized in that, The system also includes: The optimized control module, based on the reducer control results, acquires the adjusted short-time sampled torque and speed data for distribution statistics, corrects the probability centroid of the reducer control results by offset, and updates the dynamic transition window to generate the optimized control results for the reducer. The optimized control results of the speed reducer specifically include correcting the center of gravity value, rotational distribution coefficient, and updating window parameters.

6. A speed reducer control system based on chance constraints according to claim 5, characterized in that, The optimization control module is specifically: The rotation data submodule, based on the reducer control results, acquires the adjusted short-time sampled torque and speed data and performs distribution statistics, calculates the frequency ratio within the statistical interval, analyzes the joint distribution relationship between torque and speed, and obtains the rotation distribution coefficient. The center of gravity correction submodule calls the rotation distribution coefficient, compares the distribution differences of multiple intervals in the rotation distribution coefficient with the probability center of gravity position of the control result, calculates the offset of the probability center of gravity, and superimposes the offset with the probability center of gravity position in the control result to generate a corrected center of gravity value. The window update submodule performs sliding iterative updates on the corrected center of gravity value in conjunction with the dynamic transition window, records the sequence of numerical changes within the window during the iteration process, and generates the optimized control result of the reducer.

7. A speed reducer control system based on chance constraints according to claim 6, characterized in that, The sliding iterative update process specifically involves continuously iterating the corrected centroid value within the dynamic transition window according to a fixed sliding step size, and calculating the difference between the corrected centroid value of each iteration and the corrected centroid value at the previous time point during the iteration process.

Images