Temperature measurement error correction method based on hemispherical harmonic oscillator

Abstract

The invention discloses a temperature measurement error correction method based on a hemispherical resonator, and the method comprises the steps: obtaining the current measurement data of the hemispherical resonator in a current measurement environment, and uploading the current measurement data to a data storage center and a fusion neural network model; the temperature compensation multilayer perceptron unit of the fusion neural network model outputs a temperature frequency compensation value of the hemispherical harmonic oscillator, and the aging compensation long-short-term memory network unit of the fusion neural network model outputs an aging frequency compensation value of the hemispherical harmonic oscillator. A fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value and then outputs a frequency adjustment value of the hemispherical harmonic oscillator, and a compensated temperature value is obtained according to the frequency adjustment value of the hemispherical harmonic oscillator; and meanwhile, temperature frequency compensation and aging frequency compensation are carried out on the hemispherical harmonic oscillator in the temperature measurement process, so that the compensation precision of the hemispherical harmonic oscillator in the temperature measurement process is improved, and meanwhile, the measurement precision of the hemispherical harmonic oscillator can be effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of sensor measurement, and in particular to a temperature measurement error correction method based on a hemispherical resonator. Background Art

[0002] As a high-precision sensor element, the hemispherical resonator is widely used in the fields of inertial navigation and precision measurement. Its working principle is based on vibration characteristics, and physical quantity information is obtained by detecting vibration parameters. However, the performance of the hemispherical resonator is significantly affected by temperature, which seriously affects the measurement accuracy and reliability. Therefore, in the working process of the hemispherical resonator, it is necessary to compensate for the error caused by temperature in order to ensure the accuracy of the measurement of the hemispherical resonator. However, the hemispherical resonator will gradually age during use, and the hemispherical resonator will also affect the accuracy of measurement after aging. Then the existing hemispherical resonator temperature measurement error correction method only compensates for the error caused by the temperature of the hemispherical resonator, and does not consider the temperature change of the hemispherical resonator measurement environment and the aging of the hemispherical resonator caused by long-term use. If only the error caused by temperature is compensated, the aging problem of the hemispherical resonator is not considered, which is easy to cause the compensation accuracy to decrease, the error correction is also inaccurate, and also leads to the reduction of the temperature measurement accuracy of the hemispherical resonator. Summary of the invention

[0003] The present application provides a temperature measurement error correction method based on a hemispherical resonator to solve the problem of low temperature measurement accuracy of the hemispherical resonator under the influence of temperature and after aging.

[0004] In order to solve the above technical problems, in the first aspect, the present application provides a temperature measurement error correction method based on a hemispherical resonator. Acquire current measurement data of the hemispherical resonator in the current measurement environment and upload it to the data storage center and the fusion neural network model; the current measurement data includes at least one temperature value of the hemispherical resonator in the current measurement environment, the accumulated usage time of the hemispherical resonator, and the operation time of the hemispherical resonator after the last maintenance; The data storage center is used to store the current measurement data and the historical measurement data of the hemispherical resonator; The fusion neural network model is an artificial intelligence model trained based on historical measurement data; Outputting the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multilayer perceptron unit of the fusion neural network model; Outputting the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model; The fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value and outputs the temperature frequency compensation value of the hemispherical resonator; The fused neural network model obtains a compensated temperature value according to the frequency adjustment value of the hemispherical resonator.

[0005] In one embodiment, the method of obtaining the current measurement data of the hemispherical resonator in the current measurement environment and uploading it to the data storage center and the fusion neural network model includes: Obtain the temperature values ​​of different positions of the hemispherical resonator in the current measurement environment and upload them to the data storage center and the fusion neural network model; The accumulated usage time of the hemispherical resonator in the current measurement environment of the hemispherical resonator is obtained and uploaded to the data storage center and the fusion neural network model.

[0006] In one embodiment, the method of outputting the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multilayer perceptron unit of the fusion neural network model includes: Input at least one temperature value as an input node in the first input layer; Each neuron in the hidden layer receives different temperature values ​​from the first input layer, and outputs the temperature data value after weighted average summation according to preset weights to the temperature compensation output layer; The temperature compensation output layer performs weighted average summation of the temperature data values ​​and outputs a temperature frequency compensation value.

[0007] In one embodiment, the temperature data value formula obtained by weighted average summation according to preset weights is: S1=W1T1+W2T2+W3T3+······+W N T N , T1, T2, T3, T N is the temperature value, W1, W2, W3, W N T1, T2, T3, T N The corresponding weights respectively.

[0008] In one embodiment, the calculation formula for outputting the temperature frequency compensation value after weighted average summation of the temperature data values ​​is: yt=V1S1+V2S2+V3S3+······+V N S N , S1, S2, S3, S N is the temperature data value output by the hidden layer, V1, V2, V3, V N S1, S2, S3, S N The corresponding weights respectively.

[0009] In one embodiment, the method of outputting the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model includes: Input the accumulated usage time of the hemispherical resonator and / or the operation time of the hemispherical resonator after the last maintenance in the second input layer; The accumulated usage time of the hemispherical resonator and / or the operation time of the hemispherical resonator after the last maintenance are input into the LSTM layer unit as time node data; The time node data is processed by the input gate subunit, the forget gate subunit, the memory unit, and the output gate subunit in the LSTM layer unit, and outputs at least one aging output value, and outputs the aging output value to the aging output layer unit; At least one aging output value is subjected to weighted summation processing in the aging output layer unit to output an aging frequency compensation value.

[0010] In one embodiment, the time node data is processed by the LSTM layer unit through the input gate subunit, the forget gate subunit, the memory unit, and the output gate subunit to output at least one aging output value, and the method of outputting the aging output value to the aging output layer unit includes: Obtain the input gate value through the input gate subunit calculation; Obtain the forget gate value through the forget gate subunit calculation; The memory cell value is calculated by inputting the gate value and the forget gate value; The output gate subunit calculates the aging output value based on the time node data and the memory unit value.

[0011] In one embodiment, the method of outputting the aging frequency compensation value after the at least one aging output value is weighted summed in the aging output layer unit is: According to the formula y a =u1l1+u2l2+······+ u n l n ,y a is the aging frequency compensation value, l1, l2, l n is the aging output value, u1, u2, u n is l1, l2, l n The corresponding weight.

[0012] In one embodiment, at the fusion layer, the formula y = w T yt + w A y a Calculate the frequency adjustment value of the hemispherical resonator, w T and w A is the fusion layer weight.

[0013] In one of the embodiments, after the fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value and outputs the frequency adjustment value of the hemispherical resonator, it is determined whether the output frequency adjustment value of the hemispherical resonator is accurate.

[0014] In one embodiment, the method for determining whether the frequency adjustment value of the output hemispherical resonator is accurate is: Get the real resonator frequency value of the hemispherical resonator without the influence of temperature and aging; Obtain the frequency adjustment value of the hemispherical resonator output by the fusion neural network model; Determine whether the difference between the actual resonator frequency value and the corrected resonator frequency is within a preset threshold; If yes, it proves that the frequency adjustment value of the hemispherical resonator is accurate.

[0015] In a second aspect, the present application further provides a computer device, comprising a processor and a memory, wherein the memory is used to store a computer program, and when the computer program is executed by the processor, a temperature measurement error correction method based on a hemispherical resonator is implemented.

[0016] In a third aspect, the present application further provides a computer-readable storage medium storing a computer program, which implements the temperature measurement error correction method based on a hemispherical resonator when executed by a processor.

[0017] Compared with the prior art, this application has at least the following beneficial effects: The present application obtains the current measurement data of the hemispherical resonator in the current measurement environment and uploads it to the data storage center and the fusion neural network model through a temperature measurement error correction method based on a hemispherical resonator, outputs the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multi-layer perceptron unit of the fusion neural network model, outputs the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model, and the fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value to output the frequency adjustment value of the hemispherical resonator, obtains the compensated temperature value according to the frequency adjustment value of the hemispherical resonator, and simultaneously performs temperature frequency compensation and aging frequency compensation on the hemispherical resonator during the temperature measurement process, thereby improving the compensation accuracy of the hemispherical resonator in temperature measurement, and can also effectively improve the measurement accuracy of the hemispherical resonator, making the temperature measurement error smaller. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic flow chart of a temperature measurement error correction method based on a hemispherical resonator according to an embodiment of the present application; Figure 2A schematic diagram of the structure of a computer device shown in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0020] The characteristics of a hemispherical resonator are closely related to temperature and can be used for temperature measurement or to assist in temperature measurement error correction in certain situations. Hemispherical resonators are often made of high-quality quartz materials, which have an extremely low coefficient of thermal expansion, which means that when the temperature changes, the physical size and vibration characteristics of the hemispherical resonator change very little. In the field of high-precision inertial navigation and control, such as hemispherical resonator gyroscopes (HRGs), the vibration characteristics of hemispherical resonators are used to measure angular velocity. Temperature fluctuations can affect the vibration frequency and mode of the hemispherical resonator, thereby introducing errors into the gyroscope measurement results. Therefore, from this perspective, the temperature changes can be inferred by monitoring the changes in the vibration characteristics of the hemispherical resonator.

[0021] However, the prior art usually only considers that the hemispherical resonator changes in size due to thermal expansion and contraction at different temperatures, which in turn affects the vibration characteristics. By accurately measuring these changes and using appropriate algorithms, the temperature measurement error can be corrected. However, the hemispherical resonator will inevitably have aging problems during long-term use. The output signal of the aged hemispherical resonator will show greater fluctuations during long-term measurements. This is because aging causes the consistency of material properties to decrease, and the internal stress distribution may also change. Taking the application of temperature measurement as an example (inferring the temperature by monitoring the changes in the vibration characteristics of the resonator), aging causes the resonator signals measured at different times at the same temperature to have large differences, which cannot stably reflect the true temperature, affecting the stability of the measurement.

[0022] Therefore, the present application compensates for the temperature measurement value of the hemispherical resonator from the aspects of temperature and aging. The fusion neural network model can output the frequency adjustment value of the hemispherical resonator that combines the temperature frequency compensation value and the aging frequency compensation value according to the input temperature value of the hemispherical resonator and the cumulative usage time of the hemispherical resonator, and then obtain the compensated temperature value according to the frequency adjustment value of the hemispherical resonator. It not only simply considers the influence of temperature on the measurement result, but also makes the temperature value measured by the hemispherical resonator more accurate.

[0023] Please refer to Figure 1 The present application provides a temperature measurement error correction method based on a hemispherical resonator. S1, obtaining the current measurement data of the hemispherical resonator in the current measurement environment and uploading it to the data storage center and the fusion neural network model; The current measurement data includes at least one temperature value of the hemispherical resonator under the current measurement environment and the accumulated usage time of the hemispherical resonator; The data storage center is used to store the current measurement data and the historical measurement data of the hemispherical resonator; The method for obtaining the current measurement data of the hemispherical resonator in the current measurement environment and uploading it to the data storage center and the fusion neural network model includes: S11, obtaining the temperature values ​​of different positions under the current measurement environment of the hemispherical resonator, and uploading them to the data storage center and the fusion neural network model; the temperature value can be one temperature value or multiple temperature values, and the data storage center stores the temperature value for subsequent use; S12. Obtain the cumulative usage time of the hemispherical resonator in the current measurement environment of the hemispherical resonator, and upload it to the data storage center and the fusion neural network model.

[0024] The fusion neural network model is an artificial intelligence model trained based on historical measurement data; the historical measurement data includes a large number of temperature values ​​and the frequency values ​​of the hemispherical resonator output corresponding to the temperature values, and the corresponding accumulated usage time of the hemispherical resonator. The accuracy of the fusion neural network model is improved through training with a large amount of data.

[0025] S2, outputting the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multilayer perceptron unit of the fusion neural network model; The method of outputting the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multilayer perceptron unit of the fusion neural network model includes: S21, input at least one temperature value as an input node in the first input layer; for example, if there are three temperature sensors at different locations, the input layer has three nodes corresponding to the three temperature values ​​respectively; S22, each neuron in the hidden layer receives different temperature values ​​from the first input layer, and outputs the temperature data value after weighted average summation according to preset weights to the temperature compensation output layer; The hidden layer processes the input temperature value through the weighted calculation and activation function of the neuron. Each neuron receives the signal from the first input layer or the previous hidden layer, which is multiplied by the corresponding weight and then summed. Taking the temperature compensation multilayer perceptron unit with two hidden layers as an example, the temperature value of the first input layer is transmitted to the 50 neurons of the first hidden layer. For a neuron in the first hidden layer, it multiplies the temperature values ​​T1, T2, and T3 from the three nodes of the input layer (assuming three temperature sensors) by the corresponding weights W1, W2, and W3, and then sums them up.

[0026] After that, the summation result will be processed by a nonlinear activation function (such as the ReLU function). If S1 is greater than 0, the neuron outputs S1; if S1 is less than 0, it outputs 0. This output result will be used as the input signal of the next layer (the second hidden layer) of neurons. The neurons in the second hidden layer will repeat the above weighted summation and activation function processing process to further extract and transform the features in the temperature data to better fit the complex nonlinear relationship between temperature and hemispherical resonator parameters, and finally provide more valuable information for the temperature compensation output layer to calculate the temperature frequency compensation value.

[0027] The temperature data value formula obtained by weighted average summation according to the preset weights is: S1=W1T1+W2T2+W3T3+······+W N T N , T1, T2, T3, T N is the temperature value, W1, W2, W3, W N T1, T2, T3, T N The corresponding weights respectively.

[0028] S23, the temperature compensation output layer performs weighted average summation of the temperature data values ​​and outputs a temperature frequency compensation value.

[0029] The calculation formula for outputting the temperature frequency compensation value after weighted average summation of the temperature data values ​​is: yt=V1S1+V2S2+V3S3+······+V N S N , S1, S2, S3, S N is the temperature data value output by the hidden layer, V1, V2, V3, V N S1, S2, S3, S N The corresponding weights respectively.

[0030] The temperature compensation output layer has only one node, which performs a weighted summation of the values ​​of all neurons output by the last hidden layer. The weights here are continuously adjusted and determined by the back propagation algorithm during the training process. The summation result yt is the output temperature frequency compensation value, which is directly used as the final output of the temperature compensation multi-layer perceptron unit to compensate for the frequency deviation of the hemispherical resonator caused by temperature changes.

[0031] S3, outputting the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model; The method of outputting the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model comprises: S31, inputting the accumulated usage time of the hemispherical resonator and / or the operation time of the hemispherical resonator after the last maintenance in the second input layer; S32, the accumulated usage time of the hemispherical resonator and / or the operation time of the hemispherical resonator after the last maintenance are input into the LSTM layer unit as time node data; S33, after the time node data is processed by the input gate subunit, the forget gate subunit, the memory unit, and the output gate subunit in the LSTM layer unit, at least one aging output value is output, and the aging output value is output to the aging output layer unit; The method of outputting at least one aging output value after the time node data is processed by the LSTM layer unit through the input gate subunit, the forget gate subunit, the memory unit, and the output gate subunit, and outputting the aging output value to the aging output layer unit includes: S331, obtaining an input gate value through calculation by the input gate subunit; S332, obtaining a forget gate value by calculating through a forget gate subunit; S333, calculating the memory unit value by inputting the gate value and the forget gate value; S334, the output gate sub-unit calculates the aging output value according to the time node data and the memory unit value.

[0032] S34, at least one aging output value is subjected to weighted summation processing in the aging output layer unit to output an aging frequency compensation value; When the second input layer data (such as the cumulative usage time of the hemispherical resonator and the running time after the last maintenance) enters the LSTM layer, it will first be passed to the input gate subunit, forget gate subunit and output gate at the same time. For the input gate, it will calculate an input gate value based on the second input layer input data and the hidden state of the previous moment. This value determines how much information in the current input data will be added to the memory unit. Assume that the input data is x t For example, the cumulative usage time of the hemispherical resonator, the hidden state at the last moment is ht-1 , through a series of weight matrix operations, such as i t = σ(W ix x t + W ih h t-1 +b i ), where σ is the sigmoid activation function, W ix , W ih is the weight matrix, b i is the bias term, and the calculated i t It is the input gate value.

[0033] The forget gate subunit determines which old information in the memory unit will be retained. Similarly, through similar operations, such as f t =σ(W fx x t +W fh h t-1 +b f ), get the forget gate value f t The memory unit will calculate the memory unit value based on the input gate value and the forget gate value, W fx and W fh is the weight matrix, b f is the bias term of the forget gate; C t = f t ⊙C t-1 + i t ⊙tanh(W cx x t + W ch h t-1 +b c ), where ⊙ represents element-by-element multiplication, C t-1 is the state of the memory unit at the previous moment, W cx , W ch is the weight matrix, b c is the bias term.

[0034] The output gate subunit is based on the updated memory cell value C t and the current input data x t , determines the data that is finally output to the next layer. By calculating o t = σ(W ox x t + W oh h t-1 +b o ) to get the output gate value o t , W ox , W oh is the weight matrix, b o is the bias term. The final output h t= o t ⊙ tanh(C t ), h t It will be passed to the aging output layer as the aging output value of the current LSTM layer. In this way, the LSTM layer can effectively process the time-related data of the second input layer, capture the long-term dependencies therein, and provide strong support for the subsequent determination of the aging frequency compensation value.

[0035] The method of outputting the aging frequency compensation value after the at least one aging output value is weighted summed in the aging output layer unit is: Assume that the last LSTM layer outputs n eigenvalues, denoted as l1, l2, ······l n , According to the formula y a =u1l1+u2l2+······+ u n l n ,y a is the aging frequency compensation value, l1, l2, l n is the aging output value, u1, u2, u n is l1, l2, l n The corresponding weight.

[0036] S4. The fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value to output the frequency adjustment value of the hemispherical resonator.

[0037] In the fusion neural network model, the temperature frequency compensation value output by the temperature compensation multilayer perceptron unit and the aging frequency compensation value output by the aging compensation long short-term memory network unit will be passed to the fusion layer. In the fusion layer, the formula y = w T yt + w A y a The weighted summation combines the two and finally obtains the frequency adjustment value of the hemispherical resonator which comprehensively considers the temperature and aging factors. The frequency adjustment value is used to correct the original output frequency of the hemispherical resonator. The measured temperature value is deduced according to the corrected frequency value, thereby improving the temperature measurement accuracy and making the temperature measurement error smaller.

[0038] After the fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value and outputs the frequency adjustment value of the hemispherical resonator, it is determined whether the output frequency adjustment value of the hemispherical resonator is accurate; The method for judging whether the frequency adjustment value of the output hemispherical resonator is accurate is: Get the real resonator frequency value of the hemispherical resonator without the influence of temperature and aging; The frequency of the hemispherical resonator can be measured as the true value by using a high-precision frequency standard source in an ideal environment with strictly controlled temperature and zero aging factor; Obtain the frequency adjustment value of the hemispherical resonator output by the fusion neural network model; output the frequency adjustment value of the hemispherical resonator output by the fusion neural network model; Determine whether the difference between the real resonator frequency value and the corrected resonator frequency is within a preset threshold; make a difference between the real value of the frequency of the hemispherical resonator and the frequency adjustment value of the output hemispherical resonator and compare them with the preset threshold; If the difference is within the preset threshold range, it proves that the frequency adjustment value of the hemispherical resonator is accurate. If the difference is outside the preset threshold range, it proves that the frequency adjustment value of the hemispherical resonator is inaccurate. And the smaller the difference, the more accurate the frequency adjustment value of the hemispherical resonator.

[0039] The fused neural network model obtains a compensated temperature value according to the frequency adjustment value of the hemispherical resonator.

[0040] It is known that the frequency and temperature of a hemispherical resonator usually exist as F=F0(1+α△T+β△T 2 ), where F is the actual frequency, F0 is the initial frequency, α and β are material-related coefficients, and △T is the temperature change.

[0041] After the fusion layer obtains the frequency adjustment value △F of the hemispherical resonator, the temperature change △T can be inferred according to the above formula, that is, it can be obtained by solving the equation about △T. Convert it into a quadratic equation, △F=F-F0=F0(α△T+β△T 2 ), use the root-finding formula to solve △T. After obtaining △T, adjust the temperature-related parameters in the system according to the requirements of temperature compensation. For example, if the temperature value is T, then the compensated temperature value T 补 =T-△T.

[0042] In this way, the present application obtains the current measurement data of the hemispherical resonator in the current measurement environment and uploads it to the data storage center and the fusion neural network model through a temperature measurement error correction method based on a hemispherical resonator, outputs the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multi-layer perceptron unit of the fusion neural network model, outputs the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model, and the fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value to output the frequency adjustment value of the hemispherical resonator. At the same time, the hemispherical resonator is temperature compensated and aging compensated, thereby improving the compensation accuracy of the hemispherical resonator, and at the same time, it can also effectively improve the measurement accuracy of the hemispherical resonator.

[0043] Please refer to Figure 2 In a second aspect, the present application further provides a computer device 4, comprising a processor 40 and a memory 41, wherein the memory 41 is used to store a computer program 42, and when the computer program 42 is executed by the processor 40, the temperature measurement error correction method based on the hemispherical resonator is implemented.

[0044] The computer device 4 may be a computing device such as a tablet computer, a desktop computer, or a cloud server. The computer device may include but is not limited to a processor 40 and a memory 41. Those skilled in the art will appreciate that Figure 2 It is only an example of the computer device 4 and does not constitute a limitation on the computer device 4. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0045] The processor 40 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or 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, etc.

[0046] In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 4. Further, the memory 41 may also include both an internal storage unit and an external storage device of the computer device 4. The memory 41 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory 41 may also be used to temporarily store data that has been output or is to be output.

[0047] In a fourth aspect, the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the temperature measurement error correction method based on a hemispherical resonator.

[0048] An embodiment of the present application provides a computer program product. When the computer program product is run on a computer device, the computer device implements the steps in the above-mentioned method embodiments when executing the computer device.

[0049] In several embodiments provided in the present application, it is understood that each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved.

[0050] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0051] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A temperature measurement error correction method based on a hemispherical resonator, characterized in that: Acquire current measurement data of the hemispherical resonator in the current measurement environment and upload it to the data storage center and the fusion neural network model; the current measurement data includes at least one temperature value of the hemispherical resonator in the current measurement environment and the accumulated usage time of the hemispherical resonator; The data storage center is used to store the current measurement data and the historical measurement data of the hemispherical resonator; The fusion neural network model is an artificial intelligence model trained based on historical measurement data; Outputting the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multilayer perceptron unit of the fusion neural network model; Outputting the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model; The fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value and outputs the frequency adjustment value of the hemispherical resonator; The fused neural network model obtains a compensated temperature value according to the frequency adjustment value of the hemispherical resonator.

2. The temperature measurement error correction method based on the hemispherical resonator according to claim 1, characterized in that: The method for obtaining the current measurement data of the hemispherical resonator in the current measurement environment and uploading it to the data storage center and the fusion neural network model includes: Obtain the temperature values ​​of different positions of the hemispherical resonator in the current measurement environment and upload them to the data storage center and the fusion neural network model; The accumulated usage time of the hemispherical resonator in the current measurement environment of the hemispherical resonator is obtained and uploaded to the data storage center and the fusion neural network model.

3. The temperature measurement error correction method based on the hemispherical resonator according to claim 1, characterized in that: The method of outputting the temperature frequency compensation value of the hemispherical resonator through the temperature compensation multilayer perceptron unit of the fusion neural network model includes: Input at least one temperature value as an input node in the first input layer; Each neuron in the hidden layer receives different temperature values ​​from the first input layer, and outputs the temperature data value after weighted average summation according to preset weights to the temperature compensation output layer; The temperature compensation output layer performs weighted average summation of the temperature data values ​​and outputs a temperature frequency compensation value.

4. The temperature measurement error correction method based on the hemispherical resonator as claimed in claim 3, characterized in that: The temperature data value formula obtained by weighted average summation according to the preset weights is: S1=W1T1+W2T2+W3T3+······+W N T N , T1, T2, T3, T N is the temperature value, W1, W2, W3, W N T1, T2, T3, T N The corresponding weights respectively.

5. The temperature measurement error correction method based on the hemispherical resonator as claimed in claim 4, characterized in that: The calculation formula for outputting the temperature frequency compensation value after weighted average summation of the temperature data values ​​is: yt=V1S1+V2S2+V3S3+······+V N S N , S1, S2, S3, S N is the temperature data value output by the hidden layer, V1, V2, V3, V N S1, S2, S3, S N The corresponding weights respectively.

6. The temperature measurement error correction method based on the hemispherical resonator according to claim 5, characterized in that: The method of outputting the aging frequency compensation value of the hemispherical resonator through the aging compensation long short-term memory network unit of the fusion neural network model comprises: Input the accumulated usage time of the hemispherical resonator and / or the operation time of the hemispherical resonator after the last maintenance in the second input layer; The accumulated usage time of the hemispherical resonator and / or the operation time of the hemispherical resonator after the last maintenance are input into the LSTM layer unit as time node data; The time node data is processed by the input gate subunit, the forget gate subunit, the memory unit, and the output gate subunit in the LSTM layer unit, and outputs at least one aging output value, and outputs the aging output value to the aging output layer unit; At least one aging output value is subjected to weighted summation processing in the aging output layer unit to output an aging frequency compensation value.

7. The temperature measurement error correction method based on the hemispherical resonator according to claim 6, characterized in that: The method of outputting at least one aging output value after the time node data is processed by the LSTM layer unit through the input gate subunit, the forget gate subunit, the memory unit, and the output gate subunit, and outputting the aging output value to the aging output layer unit includes: Obtain the input gate value through the input gate subunit calculation; Obtain the forget gate value through the forget gate subunit calculation; The memory cell value is calculated by inputting the gate value and the forget gate value; The output gate subunit calculates the aging output value based on the time node data and the memory unit value.

8. The temperature measurement error correction method based on the hemispherical resonator according to claim 7, characterized in that: The method of outputting the aging frequency compensation value after the at least one aging output value is weighted summed in the aging output layer unit is: According to the formula y a =u1l1+u2l2+······+ u n l n ,y a is the aging frequency compensation value, l1, l2, l n is the aging output value, u1, u2, u n is l1, l2, l n The corresponding weight.

9. The temperature measurement error correction method based on the hemispherical resonator according to claim 7, characterized in that: In the fusion layer, the formula y = w T yt + w A y a Calculate the frequency adjustment value of the hemispherical resonator.

10. The temperature measurement error correction method based on a hemispherical resonator according to claim 1, characterized in that: After the fusion layer of the fusion neural network model fuses the temperature frequency compensation value and the aging frequency compensation value and outputs the frequency adjustment value of the hemispherical resonator, it is determined whether the output frequency adjustment value of the hemispherical resonator is accurate.

11. The temperature measurement error correction method based on a hemispherical resonator according to claim 10, characterized in that: The method for judging whether the frequency adjustment value of the output hemispherical resonator is accurate is: Get the real resonator frequency value of the hemispherical resonator without the influence of temperature and aging; Obtain the frequency adjustment value of the hemispherical resonator output by the fusion neural network model; Determine whether the difference between the actual resonator frequency value and the corrected resonator frequency is within a preset threshold; If yes, it proves that the frequency adjustment value of the hemispherical resonator is accurate.

12. A computer device, characterized in that: The method comprises a processor and a memory, wherein the memory is used to store a computer program, and when the computer program is executed by the processor, the temperature measurement error correction method based on a hemispherical resonator is implemented as claimed in any one of claims 1 to 11.

13. A computer-readable storage medium, characterized in that: It stores a computer program, and when the computer program is executed by a processor, the temperature measurement error correction method based on a hemispherical resonator as claimed in any one of claims 1 to 11 is implemented.

Images