Intelligent temperature control method and system of semiconductor testing machine

Abstract

The invention relates to the technical field of temperature control, and discloses an intelligent temperature control method and system of a semiconductor testing machine. The method comprises the steps of obtaining test machine structure data, a sensor position and an original temperature; combining the original temperature and the timestamp into a temperature time sequence; calculating the temperature change rate of adjacent time, and determining a temperature abrupt change point by combining the sensor position; dividing temperature regions according to the structure of the test machine and the mutation points, and calculating a region average temperature time sequence; dividing a time sub-window, fitting a temperature change curve, extracting a slope value, and determining current and historical temperature change rates; predicting a future temperature change rate by combining the prediction model, the historical rate and the iterative optimization model; and generating a temperature control instruction and sending the temperature control instruction to the controller. According to the method, refined temperature control on different functional areas of the semiconductor testing machine can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature control, and in particular to an intelligent temperature control method and system for a semiconductor tester. Background Art

[0002] Currently, temperature control in semiconductor testers is plagued by uneven temperature distribution due to the complex internal structure and significant differences in heat dissipation conditions across different areas. This unevenness can lead to temperature polarization, where some areas are too hot while others are too cold. This phenomenon not only affects the accuracy of test results but also accelerates equipment aging and shortens its service life.

[0003] In one existing technology, in order to alleviate the temperature polarization phenomenon caused by the unreasonable layout of heating elements, it is necessary to deeply analyze the polarization characteristics of the temperature field and optimize the layout and power distribution of the heating elements. By adjusting the position and power output of the heating elements, the temperature polarization phenomenon is reduced and a uniform distribution of the temperature field is achieved. This can alleviate the aging of the equipment and increase the service life of the equipment. However, in certain specific scenarios, semiconductor testers need to maintain the differences in temperature and their changes in different functional areas inside. The existing technology only adjusts and optimizes the layout and power distribution of the heating elements, without considering the differences in temperature changes in different functional areas, and cannot achieve fine temperature control. Summary of the Invention

[0004] The present invention provides an intelligent temperature control method and system for a semiconductor tester, so as to achieve refined temperature control of the semiconductor tester.

[0005] In a first aspect, in order to solve the above technical problems, the present invention provides an intelligent temperature control method for a semiconductor tester, comprising: Obtain test machine structure data, sensor location and original temperature; Combining the original temperature and the corresponding timestamp to obtain a temperature time series; Determine a temperature mutation point by calculating the temperature change rate at adjacent times in the temperature time series, and obtain the position of the temperature mutation point by combining the temperature mutation point and the sensor position; Dividing the internal space of the test machine into several temperature zones according to the test machine structure data and the location of the temperature mutation point, and calculating the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone; Dividing the average temperature time series into time subwindows, obtaining a temperature change curve within the subwindow according to the least squares fitting method, extracting a slope value of the temperature change curve, and determining a current temperature change rate and a historical temperature change rate; Combining the time series prediction model and the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtaining the optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate; A future temperature change rate at a future time is obtained according to the optimal time series prediction model, and a temperature control instruction is generated based on the future temperature change rate, and the temperature control instruction is sent to a temperature controller.

[0006] As an optional implementation, combining the original temperature and the corresponding timestamp to obtain a temperature time series includes: Combining the original temperature and the timestamp of the corresponding temperature acquisition time to obtain an original temperature time series; For the original temperature time series, pulse interference is removed by a moving average filter algorithm, and high-frequency noise is removed by a low-pass filter to obtain a filtered temperature time series; The filtered temperature time series is subjected to secondary smoothing processing using a median filter algorithm to obtain a temperature time series.

[0007] As an optional embodiment, determining the temperature mutation point by calculating the temperature change rate at adjacent times in the temperature time series includes: Calculating the temperature change rate at adjacent times in the temperature time series, and marking the point where the temperature change rate exceeds a preset change rate threshold as a temperature mutation point; The sensor position is used to find the temperature mutation point position corresponding to the temperature mutation point.

[0008] In an optional embodiment, the internal space of the tester is divided into several temperature zones according to the tester structure data and the location of the temperature mutation point, and the mean of the temperature time series in each temperature zone is calculated to obtain the average temperature time series of the temperature zone, including: A clustering algorithm is used to cluster temperature mutation points with similar locations into one category to determine a temperature zone; wherein each temperature zone has a plurality of sensors, and each sensor corresponds to a temperature time series; In each temperature region, selecting temperature values ​​from the same timestamp in all the temperature time series, and calculating an average of the temperature values ​​as the temperature average of the timestamp; The average temperature time series of the temperature region is obtained by combining the temperature mean and the timestamp corresponding to all timestamps.

[0009] As an optional embodiment, dividing the average temperature time series into time subwindows, obtaining a temperature change curve within the subwindow according to the least squares fitting method, extracting a slope value of the temperature change curve, and determining the current temperature change rate and the historical temperature change rate include: Segmenting the average temperature time series using a preset time window size to obtain multiple temperature series segments; Applying the least squares method to fit the temperature variation curve to the temperature sequence segment; Calculating the slope value of the temperature change curve corresponding to the historical time window to obtain the historical temperature change rate; The slope value of the temperature change curve corresponding to the current time window is calculated to obtain the current temperature change rate.

[0010] As an optional embodiment, combining the time series prediction model and the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtaining the optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate, includes: Predicting the predicted temperature change rate corresponding to the current time window based on the time series prediction model and the historical temperature change rate; Adjusting the size of the time subwindow and replacing the least squares fitting function in real time by a back propagation algorithm so that the square of the difference between the predicted temperature change rate and the current temperature change rate decreases; Repeat the steps of adjusting the size of the time subwindow and replacing the least squares fitting function. When the number of iterations is greater than or equal to the preset number of iterations, or the square of the difference between the predicted temperature change rate and the current temperature change rate is less than a preset loss threshold, output the current time subwindow and the current least squares fitting function to obtain the optimal time series prediction model.

[0011] As an optional embodiment, before calculating the mean of the temperature time series in each temperature region to obtain the average temperature time series of the temperature region, the method further includes: Determine a first temperature time series corresponding to a boundary sensor of a first temperature zone, and determine a second temperature time series corresponding to a boundary sensor of a second temperature zone; wherein each temperature zone has a plurality of sensors, each sensor corresponds to a temperature time series; and the first temperature zone and the second temperature zone are two adjacent temperature zones; Calculating a gradient value between the first temperature time series and the second temperature time series; When the gradient value is greater than a preset gradient threshold, the first temperature time series and the second temperature time series are fused by weighted averaging to obtain a boundary temperature time series; The first temperature time series and the second temperature time series are replaced by the boundary temperature time series.

[0012] In a second aspect, the present invention provides an intelligent temperature control system for a semiconductor tester, comprising: Data acquisition module, used to obtain test machine structure data, sensor position and original temperature; A data preprocessing module, configured to combine the original temperature and the corresponding timestamp to obtain a temperature time series; a temperature zone division module, configured to determine a temperature mutation point by calculating the temperature change rate between adjacent times in the temperature time series, obtain the location of the temperature mutation point by combining the temperature mutation point and the sensor position, and divide the internal space of the tester into a plurality of temperature zones according to the tester structure data and the location of the temperature mutation point; An average temperature time series calculation module is used to calculate the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone; a temperature change rate calculation module, configured to divide the average temperature time series into time subwindows, obtain a temperature change curve within the subwindow using a least squares fitting method, extract a slope value from the temperature change curve, and determine a current temperature change rate and a historical temperature change rate; an optimization time series prediction model module for combining the time series prediction model with the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtaining an optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate; The result output module is used to obtain the future temperature change rate at a future time according to the optimal time series prediction model, generate a temperature control instruction based on the future temperature change rate, and send the temperature control instruction to the temperature controller.

[0013] In a third aspect, the present invention also provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the intelligent temperature control method for a semiconductor testing machine described in any one of the above is implemented.

[0014] In a fourth aspect, the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned intelligent temperature control methods for a semiconductor testing machine.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention discloses an intelligent temperature control method for a semiconductor tester, comprising the following steps: obtaining tester structural data, sensor positions, and original temperatures; combining the original temperatures and corresponding timestamps to obtain a temperature time series; determining a temperature mutation point by calculating the adjacent time temperature change rate in the temperature time series, and obtaining the position of the temperature mutation point by combining the temperature mutation point and the sensor position; dividing the internal space of the tester into a plurality of temperature zones according to the tester structural data and the temperature mutation point positions, and calculating the mean of the temperature time series in each temperature zone to obtain an average temperature time series of the temperature zone; and performing a temperature analysis on the average temperature time series. The time is divided into sub-windows, and a temperature change curve within the sub-window is obtained by fitting the least squares method. The slope value of the temperature change curve is extracted to determine the current temperature change rate and the historical temperature change rate. The time series prediction model and the historical temperature change rate are combined to obtain the predicted temperature change rate for the current time, and the optimal time series prediction model is iteratively obtained by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate. The future temperature change rate at a future time is obtained according to the optimal time series prediction model, and a temperature control instruction is generated based on the future temperature change rate, and the temperature control instruction is sent to a temperature controller.

[0016] The method divides the interior of a semiconductor tester into multiple independent temperature zones and conducts in-depth analysis and modeling of the temperature characteristics of each zone, achieving accurate predictions of the future temperature change rates of different temperature zones. This process is based on a time series prediction model, and the model parameters are continuously adjusted through iterative optimization to ensure the accuracy and reliability of the prediction results. Specifically, the method first divides the equipment into several temperature zones with similar temperature characteristics based on the internal structure and sensor layout of the tester. Then, based on the historical temperature data of each zone, a time series prediction model is constructed and trained to capture the dynamic patterns of temperature changes. Through continuous iterative optimization, the model can automatically adjust its parameters to adapt to changes in the temperature field, thereby improving prediction accuracy. Ultimately, based on these prediction results, the system can develop a more refined temperature control strategy to achieve precise temperature regulation of different temperature zones within the semiconductor tester, thereby achieving refined management of different zones within the semiconductor tester and providing a stable and reliable temperature environment for the semiconductor testing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of an intelligent temperature control method for a semiconductor tester provided by an embodiment of the present invention; Figure 2 The figure is a schematic structural diagram of an intelligent temperature control system for a semiconductor tester provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0019] Reference Figure 1 A first embodiment of the present invention provides an intelligent temperature control method for a semiconductor tester, comprising the following steps: S1, obtain the test machine structure data, sensor position and original temperature; S2, combining the original temperature and the corresponding timestamp to obtain a temperature time series; S3, determining a temperature mutation point by calculating a temperature change rate at adjacent times in the temperature time series, and obtaining a position of the temperature mutation point by combining the temperature mutation point and the sensor position; S4, dividing the internal space of the test machine into a plurality of temperature zones according to the test machine structure data and the location of the temperature mutation point, and calculating the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone; S5, dividing the average temperature time series into time subwindows, obtaining a temperature change curve within the subwindow by least squares fitting, extracting a slope value of the temperature change curve, and determining a current temperature change rate and a historical temperature change rate; S6, combining the time series prediction model and the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtaining an optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate; S7, obtaining a future temperature change rate at a future time according to the optimal time series prediction model, generating a temperature control instruction based on the future temperature change rate, and sending the temperature control instruction to a temperature controller.

[0020] In step S1 , the test machine structure data, sensor positions and original temperatures are obtained.

[0021] It should be noted that the test machine structural data refers to the location information of the test machine's key cooling and heating components. In this embodiment, a three-dimensional rectangular coordinate system is first established for the test machine. The key cooling and heating components of the test machine are the heating element and the cooling system air outlet. A temperature sensor array is deployed according to the coordinate positions of the key cooling and heating components. The deployment coordinate position of each temperature sensor is recorded, and the sampling parameters of the temperature sensor array are set to a sampling frequency greater than or equal to 1Hz. The temperature acquisition process is then initiated to obtain the raw temperature.

[0022] In step S2, the original temperature and the corresponding time stamp are combined to obtain a temperature time series.

[0023] It should be noted that the timestamp means that when the original temperature is collected by the temperature sensor and transmitted to the database, the database will add a mark to each original temperature, indicating the time when each original temperature is collected. This mark is the timestamp.

[0024] The step S2 of combining the original temperature and the corresponding timestamp to obtain a temperature time series includes: S21, combining the original temperature and the timestamp of the corresponding temperature acquisition time to obtain an original temperature time series; S22, for the original temperature time series, removing pulse interference by a moving average filtering algorithm, and removing high-frequency noise by a low-pass filter, to obtain a filtered temperature time series; S23, performing secondary smoothing processing on the filtered temperature time series using a median filtering algorithm to obtain a temperature time series.

[0025] It should be noted that the raw temperature time series is obtained by combining the raw temperatures and corresponding timestamps in the database and sorting them in the order of the timestamps. Impulse interference is a transient, large-amplitude noise signal that typically appears as spikes or sudden changes in the raw temperature time series. It may be caused by external electromagnetic interference, equipment failure, or sudden errors in signal transmission. High-frequency noise is a rapidly changing signal that typically appears as high-frequency fluctuations in the raw temperature time series. It may be caused by electromagnetic interference, sensor noise, or high-frequency interference in signal transmission. A low-pass filter allows low-frequency signals to pass while suppressing high-frequency signals. It uses mathematical operations to weaken high-frequency components while retaining the low-frequency portion of the signal, thereby removing high-frequency noise. A median filter is a nonlinear filtering method that replaces the value of each data point in the series with the median value of its neighborhood. It is particularly suitable for removing impulsive noise (such as salt and pepper noise) because it is insensitive to extreme values ​​(noise points).

[0026] In step S3, the temperature mutation point is determined by calculating the temperature change rate of adjacent times in the temperature time series, including: S31, calculating the temperature change rate of adjacent time periods in the temperature time series, and marking the point where the temperature change rate exceeds a preset change rate threshold as a temperature mutation point; S32: Using the sensor position, find the temperature mutation point position corresponding to the temperature mutation point.

[0027] It's important to note that the preset rate-of-change threshold is a key parameter, defining the maximum acceptable temperature change within the temperature sensor's acquisition interval. The selection of the temperature rate-of-change threshold requires comprehensive consideration of multiple factors, including equipment characteristics, process requirements, sensor accuracy, historical data, actual operating conditions, and experimental verification. By properly setting and dynamically adjusting the threshold, precise monitoring and control of temperature changes can be achieved, improving system reliability and stability.

[0028] In a feasible embodiment, for a semiconductor testing machine, if the process requires that the temperature change does not exceed ±0.5℃ / min; if historical data shows that the average value of the temperature change rate is ±0.3℃ / min and the standard deviation is ±0.2℃ / min, the threshold can be set to ±0.5℃ / min; in the equipment startup phase, the temperature change rate threshold can be set to ±2℃ / min; in the stable operation phase, the threshold can be set to ±0.5℃ / min, etc.

[0029] In one feasible embodiment, the temperature change rate between each time point and its adjacent time points is calculated. This rate of change reflects the dynamic changes in temperature over time. If the temperature change rate at a certain time point exceeds a pre-set rate of change threshold, then this time point will be marked as a temperature mutation point. The setting of this threshold is based on an in-depth analysis of the temperature field variation characteristics, and aims to distinguish normal temperature fluctuations from significant mutation events that may affect the division of temperature zones. After identifying the temperature mutation points, the spatial position information of the sensor is further used to determine the specific location of these mutation points in the semiconductor tester.

[0030] In step S4, the internal space of the tester is divided into several temperature zones according to the tester structure data and the location of the temperature mutation point, and the mean of the temperature time series in each temperature zone is calculated to obtain the average temperature time series of the temperature zone, including: S41, using a clustering algorithm to cluster temperature mutation points with similar locations into one category to determine a temperature region; wherein each temperature region has a plurality of sensors, and each sensor corresponds to a temperature time series; S42, in each temperature region, selecting temperature values ​​from the same timestamp in all the temperature time series, and calculating an average of the temperature values ​​as the temperature average of the timestamp; S43, combining the temperature mean values ​​and the timestamps corresponding to all the timestamps to obtain an average temperature time series of the temperature region.

[0031] It should be noted that the division of temperature zones is a key strategy that can significantly improve the accuracy and efficiency of temperature control. By dividing the internal space of the test machine into multiple temperature zones, each zone can be independently temperature-controlled according to its own temperature change characteristics. This refined management method can more accurately meet the temperature requirements of different zones; it can reduce errors caused by global control; and it can help quickly locate problem areas and reduce troubleshooting time. Within the temperature zone, the temperature change rate is usually lower than the preset threshold, the temperature change is relatively stable, and there will be no drastic temperature fluctuations; the temperature mean within each temperature zone is relatively stable and significantly different from the temperature mean of other zones.

[0032] In this embodiment of the present invention, the K-means clustering algorithm is selected. Based on the location of the temperature mutation points, temperature mutation points within the test machine are clustered into one category if they are close in location. Temperature mutation points of the same type serve as the boundaries of the temperature zones, dividing the test machine into K temperature zones. For each temperature zone, the temperature-time series corresponding to all temperature sensors within the same temperature zone are used and arranged in order of position to form a two-dimensional matrix. The rows represent the temperature-time series corresponding to a particular temperature sensor, and the columns represent the timestamp (the moment the temperature was collected). The columns of the two-dimensional matrix are averaged to obtain a vector representing the average temperature-time series for the temperature zone. The formula is as follows: Assuming that there are N temperature sensors in the temperature area and each sensor collects temperature data with M timestamps, we can get a two-dimensional matrix : Indicates the A temperature sensor in the The matrix rows represent the temperature time series corresponding to the temperature sensor; the matrix columns represent the temperature values ​​of the sensor at the same time stamp.

[0033] Pair Matrix Calculate the mean of each column to obtain the vector representing the average temperature time series of the temperature area : Among them, for example, The calculation formula is: in, express Matrix The temperature value of row 1 is in column 1. The calculation formulas for other elements are similar and will not be repeated here.

[0034] In step S5, the average temperature time series is divided into time sub-windows, a temperature change curve within the sub-window is obtained by least squares fitting, a slope value is extracted from the temperature change curve, and the current temperature change rate and the historical temperature change rate are determined, including: S51, segmenting the average temperature time series using a preset time window size to obtain multiple temperature series segments; S52, applying the least square method to fit the temperature variation curve to the temperature sequence segment; S53, calculating the slope value of the temperature change curve corresponding to the historical time window to obtain the historical temperature change rate; S54, calculating the slope value of the temperature change curve corresponding to the current time window to obtain the current temperature change rate.

[0035] It should be noted that the preset time window size is a key parameter for obtaining the historical and current temperature change rates. Selecting an appropriate time window size is crucial for ensuring data smoothness, accurately reflecting the temperature change rate, and responding to changes in a timely manner. The present invention presets the time window size based on actual application requirements.

[0036] In one feasible embodiment, if a rapid response to temperature changes is required, the time window is set to a smaller value; if the goal is to analyze long-term trends, the time window should be larger. For example, in real-time temperature control, the time window can be set to 10 to 30 seconds; in daily average temperature analysis, the time window can be set to 1 to 6 hours.

[0037] For example, a preset time window size is used to average the temperature time series. Segmentation is performed to obtain multiple temperature sequence segments. For example, the temperature sequence segments can be expressed as: in, 、 、 Represent the 1st, 2nd, and mth temperature sequence fragments respectively.

[0038] In an embodiment of the present invention, the temperature variation curve is fitted to the temperature sequence segments using the least squares method, wherein a linear fitting function is used to fit the temperature variation curve of each temperature sequence segment. The least squares method finds optimal model parameters by minimizing the sum of squared errors between observed data and model predictions, and then fits the temperature variation curve of the temperature sequence segments to extract the slope value of each temperature sequence segment.

[0039] In this embodiment of the present invention, the historical temperature sequence segment set , current temperature sequence fragment collection Defined as: in, represents a set of historical temperature sequence fragments, Represents the current temperature sequence fragment set, 、 、 Represent the kth, n-1th, and nth temperature sequence fragments respectively.

[0040] By collecting historical temperature series fragments , current temperature sequence fragment collection The historical temperature change rate can be obtained by fitting the change curve using the least squares method and extracting the slope value. and the current temperature change rate .

[0041] For example, Indicates the use of least squares method from The calculation methods of other elements are similar and will not be described here.

[0042] In step S6, the time series prediction model and the historical temperature change rate are combined to obtain the predicted temperature change rate for the current time, and the optimal time series prediction model is iteratively obtained by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate.

[0043] It should be noted that the time series forecasting model is initialized based on the historical and current temperature change rates, combined with an existing regression model. Commonly used models include the Autoregressive Moving Average (ARMA) model and the Seasonal Autoregressive Integrated Moving Average (SARIMA) model. These models can capture the periodicity, trend, and randomness of the temperature change rate. Using this constructed time series model, the temperature change rate can be predicted for future time subwindows.

[0044] In step S6, the time series prediction model and the historical temperature change rate are combined to obtain a predicted temperature change rate for the current time, and the optimal time series prediction model is iteratively obtained by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate, including: S61, predicting the predicted temperature change rate corresponding to the current time window based on the time series prediction model and the historical temperature change rate; S62, adjusting the size of the time subwindow and replacing the least squares fitting function in real time by a back propagation algorithm, so that the square of the difference between the predicted temperature change rate and the current temperature change rate decreases; S63, repeat the steps of adjusting the size of the time subwindow and replacing the least squares fitting function. When the number of iterations is greater than or equal to the preset number of iterations, or the square of the difference between the predicted temperature change rate and the current temperature change rate is less than a preset loss threshold, output the current time subwindow and the current least squares fitting function to obtain the optimal time series prediction model.

[0045] In a feasible embodiment, an ARMA model is used as a time series model, the parameters of the time series model are initialized according to the historical temperature change rate, and the current temperature change rate is predicted using the ARMA time series model. The ARMA model formula is as follows: in, is the historical temperature change rate in The value of the moment; is the historical temperature change rate in The value of the moment; is a constant term; is the order of the autoregressive part; is the order of the moving average part; is the autoregressive part of parameters; is the zth parameter of the moving average part; It is The error term between the predicted value and the true value at the moment; It is The error term between the predicted value and the true value at the moment.

[0046] In one possible embodiment, the default ( ∈[1, q]) takes random numbers according to the white noise probability density distribution, and uses maximum likelihood estimation (MLE) based on the historical temperature change rate to find the parameter value that maximizes the likelihood function (the corresponding likelihood function is the probability of observing data under given parameters) and uses it as the default parameter value.

[0047] In a feasible embodiment, the initialized ARMA time series prediction model is used to predict the current temperature change rate, and the difference between the predicted current temperature change rate and the current temperature change rate is calculated as , calculate the square of the difference between the predicted current temperature change rate and the current temperature change rate, obtain the L2 loss value, and use the back propagation algorithm to adjust 、 , time window size w, least squares fitting function, etc., minimize the L2 loss value, and iterate to obtain the optimal time series model. The calculation formula is as follows: in, Represents the new autoregressive part after iteration parameters, Represents the zth parameter of the new moving average part after iteration, Indicates the new time window size after iteration, is the learning rate, which controls the step size of parameter updates.

[0048] As the temperature is collected, the new temperature and the corresponding timestamp will be entered into the average temperature time series, thereby updating the current temperature change rate and the historical temperature change rate. It then enters the historical temperature change rate, represents the error between the predicted value and the true value of the historical temperature change rate, and participates in the ARMA model's prediction and parameter iteration process of the temperature change rate in the future.

[0049] In a feasible embodiment, the initialized ARMA time series prediction model is pre-trained and can be directly imported from the memory.

[0050] In step S7, the intelligent temperature control method for a semiconductor tester according to claim 1 is characterized in that, before calculating the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone, the method further comprises: S71, determining a first temperature-time sequence corresponding to a boundary sensor of a first temperature zone, and determining a second temperature-time sequence corresponding to a boundary sensor of a second temperature zone; wherein each temperature zone has a plurality of sensors, each sensor corresponding to a temperature-time sequence; and the first temperature zone and the second temperature zone are two adjacent temperature zones; S72, calculating the gradient value of the first temperature time series and the second temperature time series; S73, when the gradient value is greater than a preset gradient threshold, fusing the first temperature time series and the second temperature time series by weighted averaging to obtain a boundary temperature time series; S74: Replace the first temperature time series and the second temperature time series with the boundary temperature time series.

[0051] It should be noted that in temperature control and monitoring systems, especially those involving multiple temperature zones, discontinuities may occur in temperature data at zone boundaries. This discontinuity can be caused by a variety of factors, such as differences in sensor accuracy, inaccurate zone demarcation, and the influence of localized heat or cooling sources. To prevent this discontinuity from negatively impacting temperature control and monitoring results, it is necessary to fuse temperature data from adjacent zones. The boundaries of temperature zones are determined by identifying temperature discontinuities in the temperature field. Temperature discontinuities are specific locations in the temperature sensor array where the rate of temperature change exceeds a preset threshold. These points typically mark significant differences between zones in the temperature field and serve as the basis for demarcating temperature zone boundaries. After identifying the temperature discontinuities, boundary sensors are defined based on these discontinuities and the sensors surrounding them. Specifically, boundary sensors include the temperature discontinuity itself, as well as other temperature sensors within a sensor array distance of no more than one sensor spacing. "No more than one sensor spacing" here means that these sensors are adjacent to the discontinuity in the sensor array layout, meaning they are either directly connected to the discontinuity or located within the same local area as the discontinuity, but no more than one sensor spacing away. The gradient value refers to the rate of change of the first temperature time series and the second temperature time series at the same time with respect to the spatial distance. The formula is as follows: in, 、 Represents the temperature of the n1th and n2th temperature sensors at the jth time stamp, Indicates the location of the n1th and n2th temperature sensors, It represents the gradient of the temperature between the n1th and n2th temperature sensors with respect to the distance at the jth time stamp.

[0052] It should be noted that the gradient threshold is a key parameter that determines temperature convergence. It is selected based on the device structure and material properties of the semiconductor tester. In the embodiment of the present invention, the gradient threshold is selected to be 5°C / m.

[0053] In one feasible embodiment, a weighted average algorithm is used to fuse the first and second temperature time series. The weighted average algorithm is a simple and effective method that calculates the fused temperature value by assigning a weight to each sensor data point. The weight is dynamically adjusted based on the sensor's accuracy, location, or data quality. The formula is as follows: 、 Represents the temperature of the n1th and n2th temperature sensors at the jth time stamp, and are weights, Indicates the temperature of the boundary temperature sensor after fusion at the jth time stamp. Regarding the choice of weight, if the accuracy of sensor 1 is higher, a larger weight can be given .

[0054] To facilitate understanding of the present invention, some preferred embodiments of the present invention are further described below.

[0055] In this embodiment, the intelligent temperature control of the semiconductor tester involves temperature zone division and a temperature change rate time series prediction model for more refined control of the internal temperature of the semiconductor tester.

[0056] In order to make corresponding operations to control the temperature change rate in real time based on the temperature change rate time series prediction model for the future temperature change rate prediction, the following two implementation methods are provided: wherein, the computer and the controller are pre-connected via a wired network.

[0057] In one embodiment, the temperature change rate time series prediction model sends the predicted temperature change rate in the future to the controller via the industrial Ethernet configured on the controller, so that the controller executes corresponding control measures based on the received data and notifies the operator to take measures; In another embodiment, the temperature change rate time series prediction model sends the predicted temperature change rate in the future to the controller through a communication module configured on the computer, so that the control end executes corresponding control measures based on the received data and notifies the operator to take emergency measures; wherein, the computer and the controller are pre-connected via a wireless network.

[0058] Reference Figure 2 A second embodiment of the present invention provides an intelligent temperature control system for a semiconductor tester, comprising: Data acquisition module, used to obtain test machine structure data, sensor position and original temperature; A data preprocessing module, configured to combine the original temperature and the corresponding timestamp to obtain a temperature time series; a temperature zone division module, configured to determine a temperature mutation point by calculating the temperature change rate between adjacent times in the temperature time series, obtain the location of the temperature mutation point by combining the temperature mutation point and the sensor position, and divide the internal space of the tester into a plurality of temperature zones according to the tester structure data and the location of the temperature mutation point; An average temperature time series calculation module is used to calculate the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone; a temperature change rate calculation module, configured to divide the average temperature time series into time subwindows, obtain a temperature change curve within the subwindow using a least squares fitting method, extract a slope value from the temperature change curve, and determine a current temperature change rate and a historical temperature change rate; an optimization time series prediction model module for combining the time series prediction model with the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtaining an optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate; The result output module is used to obtain the future temperature change rate at a future time according to the optimal time series prediction model, generate a temperature control instruction based on the future temperature change rate, and send the temperature control instruction to the temperature controller.

[0059] In summary, the present invention discloses an intelligent temperature control method for a semiconductor tester, comprising: obtaining tester structural data, sensor position and original temperature; combining the original temperature and the corresponding timestamp to obtain a temperature time series; determining a temperature mutation point by calculating the adjacent time temperature change rate in the temperature time series, and obtaining the position of the temperature mutation point by combining the temperature mutation point and the sensor position; dividing the internal space of the tester into several temperature zones according to the tester structural data and the temperature mutation point position, and calculating the mean of the temperature time series in each temperature zone to obtain an average temperature time series of the temperature zone; and calculating the average temperature time series of the average temperature time series. The time series is divided into time sub-windows, and a temperature change curve within the sub-window is obtained by fitting the least squares method. The slope value of the temperature change curve is extracted to determine the current temperature change rate and the historical temperature change rate; the time series prediction model and the historical temperature change rate are combined to obtain the predicted temperature change rate for the current time, and the optimal time series prediction model is iteratively obtained by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate; the future temperature change rate at a future time is obtained based on the optimal time series prediction model, and a temperature control instruction is generated based on the future temperature change rate, and the temperature control instruction is sent to the temperature controller. The method divides the temperature zone inside the semiconductor tester and iteratively optimizes the time series prediction model for each temperature zone to predict the future temperature change rate of different temperature zones, thereby achieving the purpose of finely controlling the temperature in different temperature zones within the semiconductor tester.

[0060] It should be noted that the intelligent temperature control system of a semiconductor testing machine provided in an embodiment of the present invention is used to execute all process steps of the intelligent temperature control method of a semiconductor testing machine in the above embodiment. The working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0061] An embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an intelligent temperature control program for a semiconductor tester. When the processor executes the computer program, the steps of the above-mentioned intelligent temperature control method for a semiconductor tester are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of the modules in the above-mentioned device embodiments are realized, such as the temperature change rate calculation module.

[0062] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0063] The electronic device may be a computing device such as a desktop computer, notebook, PDA, or smart tablet. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will appreciate that the aforementioned components are merely examples of electronic devices and do not constitute a limitation of the electronic device. The electronic device may include more or fewer components than those described above, or a combination of certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, and the like.

[0064] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor is the control center of the electronic device and connects various parts of the entire electronic device using various interfaces and lines.

[0065] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and accessing the data stored in the memory. The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function or an image playback function); the data storage area may store data generated based on the use of the mobile phone (such as audio data, a phone book, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0066] If the module / unit integrated into the electronic device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium can be appropriately increased or decreased based on the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, based on legislation and patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0067] It should be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive effort.

[0068] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. An intelligent temperature control method for a semiconductor tester, characterized in that: include: Obtain test machine structure data, sensor location and original temperature; Combining the original temperature and the corresponding timestamp to obtain a temperature time series; Determine a temperature mutation point by calculating the temperature change rate at adjacent times in the temperature time series, and obtain the position of the temperature mutation point by combining the temperature mutation point and the sensor position; Dividing the internal space of the test machine into several temperature zones according to the test machine structure data and the location of the temperature mutation point, and calculating the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone; Dividing the average temperature time series into time subwindows, obtaining a temperature change curve within the subwindow according to the least squares fitting method, extracting a slope value of the temperature change curve, and determining a current temperature change rate and a historical temperature change rate; Combining the time series prediction model and the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtaining the optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate; A future temperature change rate at a future time is obtained according to the optimal time series prediction model, and a temperature control instruction is generated based on the future temperature change rate, and the temperature control instruction is sent to a temperature controller.

2. The intelligent temperature control method for a semiconductor tester according to claim 1, wherein: Combining the original temperature and the corresponding timestamp to obtain a temperature time series includes: Combining the original temperature and the timestamp of the corresponding temperature acquisition time to obtain an original temperature time series; For the original temperature time series, pulse interference is removed by a moving average filter algorithm, and high-frequency noise is removed by a low-pass filter to obtain a filtered temperature time series; The filtered temperature time series is subjected to secondary smoothing processing using a median filter algorithm to obtain a temperature time series.

3. The intelligent temperature control method for a semiconductor tester according to claim 1, wherein: The determining of the temperature mutation point by calculating the temperature change rate at adjacent times in the temperature time series includes: Calculating the temperature change rate at adjacent times in the temperature time series, and marking the point where the temperature change rate exceeds a preset change rate threshold as a temperature mutation point; The sensor position is used to find the temperature mutation point position corresponding to the temperature mutation point.

4. The intelligent temperature control method for a semiconductor tester according to claim 1, wherein: The method of dividing the internal space of the test machine into a plurality of temperature zones according to the test machine structure data and the position of the temperature mutation point, and calculating the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone includes: A clustering algorithm is used to cluster temperature mutation points with similar locations into one category to determine a temperature zone; wherein each temperature zone has a plurality of sensors, and each sensor corresponds to a temperature time series; In each temperature region, selecting temperature values ​​from the same timestamp in all the temperature time series, and calculating an average of the temperature values ​​as the temperature average of the timestamp; The average temperature time series of the temperature region is obtained by combining the temperature mean and the timestamp corresponding to all timestamps.

5. The intelligent temperature control method for a semiconductor tester according to claim 1, wherein: The method of dividing the average temperature time series into time sub-windows, obtaining a temperature change curve within the sub-window by least squares fitting, extracting a slope value from the temperature change curve, and determining a current temperature change rate and a historical temperature change rate comprises: Segmenting the average temperature time series using a preset time window size to obtain multiple temperature series segments; Applying the least squares method to fit the temperature variation curve to the temperature sequence segment; Calculating the slope value of the temperature change curve corresponding to the historical time window to obtain the historical temperature change rate; The slope value of the temperature change curve corresponding to the current time window is calculated to obtain the current temperature change rate.

6. The intelligent temperature control method for a semiconductor tester according to claim 1, wherein: The method combines the time series prediction model and the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtains the optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate, including: Predicting the predicted temperature change rate corresponding to the current time window based on the time series prediction model and the historical temperature change rate; Adjusting the size of the time subwindow and replacing the least squares fitting function in real time by a back propagation algorithm so that the square of the difference between the predicted temperature change rate and the current temperature change rate decreases; Repeat the steps of adjusting the size of the time subwindow and replacing the least squares fitting function. When the number of iterations is greater than or equal to the preset number of iterations, or the square of the difference between the predicted temperature change rate and the current temperature change rate is less than a preset loss threshold, output the current time subwindow and the current least squares fitting function to obtain the optimal time series prediction model.

7. The intelligent temperature control method for a semiconductor tester according to claim 1, wherein: Before calculating the temperature time series mean in each temperature region to obtain the average temperature time series of the temperature region, the method further includes: Determine a first temperature time series corresponding to a boundary sensor of a first temperature zone, and determine a second temperature time series corresponding to a boundary sensor of a second temperature zone; wherein each temperature zone has a plurality of sensors, each sensor corresponds to a temperature time series; and the first temperature zone and the second temperature zone are two adjacent temperature zones; Calculating a gradient value between the first temperature time series and the second temperature time series; When the gradient value is greater than a preset gradient threshold, the first temperature time series and the second temperature time series are fused by weighted averaging to obtain a boundary temperature time series; The first temperature time series and the second temperature time series are replaced by the boundary temperature time series.

8. An intelligent temperature control system for a semiconductor tester, characterized in that: include: Data acquisition module, used to obtain test machine structure data, sensor position and original temperature; A data preprocessing module, configured to combine the original temperature and the corresponding timestamp to obtain a temperature time series; a temperature zone division module, configured to determine a temperature mutation point by calculating the temperature change rate between adjacent times in the temperature time series, obtain the location of the temperature mutation point by combining the temperature mutation point and the sensor position, and divide the internal space of the tester into a plurality of temperature zones according to the tester structure data and the location of the temperature mutation point; An average temperature time series calculation module is used to calculate the mean of the temperature time series in each temperature zone to obtain the average temperature time series of the temperature zone; a temperature change rate calculation module, configured to divide the average temperature time series into time subwindows, obtain a temperature change curve within the subwindow using a least squares fitting method, extract a slope value from the temperature change curve, and determine a current temperature change rate and a historical temperature change rate; an optimization time series prediction model module for combining the time series prediction model with the historical temperature change rate to obtain a predicted temperature change rate for the current time, and iteratively obtaining an optimal time series prediction model by minimizing the square of the difference between the predicted temperature change rate and the current temperature change rate; The result output module is used to obtain the future temperature change rate at a future time according to the optimal time series prediction model, generate a temperature control instruction based on the future temperature change rate, and send the temperature control instruction to the temperature controller.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the intelligent temperature control method for a semiconductor testing machine according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the intelligent temperature control method for a semiconductor testing machine according to any one of claims 1 to 7.

Images