CT system temperature control method and system based on long and short memory neural network

Abstract

The invention discloses a CT system temperature control method and system based on a long and short memory neural network, and relates to the technical field of artificial intelligence, and the method comprises the steps: collecting temperature data samples of an X-ray detector under different conditions, and inputting the temperature data samples into a pre-constructed temperature control model for training, so as to obtain an optimal temperature control model; deploying the optimal temperature control model into an X-ray detector temperature control system; acquiring to-be-adjusted temperature data of the X-ray detector based on an X-ray detector temperature control system, and inputting the to-be-adjusted temperature data into the optimal temperature control model to obtain optimal temperature control input parameters; and inputting the optimal temperature control input parameter into the corresponding temperature control module to realize temperature control of the CT system, thereby being beneficial to solving the problem that the temperature of the CT system cannot be accurately controlled in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a CT system temperature control method and system based on a long short-term memory neural network. BACKGROUND

[0002] In a CT system, the X-ray detector is a core imaging component, and its temperature stability directly determines the image quality. The detector is required to be stably operated at about 40 DEG C for a long time in the clinic, and the temperature deviation needs to be controlled within ± 0.5 DEG C, otherwise it will cause the detector sensitivity drift, noise increase, image artifacts or resolution decline, and interfere with the diagnostic accuracy.

[0003] The current mainstream temperature control method mainly uses PID control or a simple adaptive algorithm. Such technology relies on manual parameter setting and belongs to lagging feedback adjustment, and is difficult to cope with complex dynamic scenarios: for example, the detector itself heats up when the CT system is continuously scanned under high load, the environmental temperature of the machine room fluctuates seasonally, and the heat dissipation requirements are different under different patient scanning protocols, etc. Temperature overshoot or response delay is easy to occur.

[0004] In addition, the detector temperature is affected by the inlet temperature, the water cooling system state, the fan speed and other variables. The fitting ability of the traditional algorithm for nonlinear relationships is limited, and the control accuracy is greatly reduced under complex working conditions.

[0005] Therefore, there is an urgent need for a method that can meet the high-precision temperature control requirements of the X-ray detector. SUMMARY

[0006] Therefore, the present application provides a CT system temperature control method and system based on a long short-term memory neural network, which can accurately control the temperature of the CT system.

[0007] To achieve the above purpose, the present application provides the following technical solutions:

[0008] A CT system temperature control method based on a long short-term memory neural network, comprising:

[0009] Collecting temperature data samples of the X-ray detector under different conditions and then inputting them into a pre-constructed temperature control model for training to obtain an optimal temperature control model;

[0010] The temperature control model includes an input layer, a hidden layer and an output layer. The input layer includes an inlet temperature node, a detector environmental temperature node, a detector surface temperature node, a CT system output temperature node and a detector target temperature node. The output layer includes an inlet fan speed node, an outlet fan speed node, a detector water cooling flow rate node and a heating belt target temperature node. The number of hidden layer nodes is determined based on the number of input layer nodes, the number of output layer nodes and an adjustment constant.

[0011] deploy the optimal temperature control model into an X-ray detector temperature control system;

[0012] collect temperature data to be adjusted of the X-ray detector based on the X-ray detector temperature control system, input the temperature data to be adjusted into the optimal temperature control model to obtain optimal temperature control input parameters;

[0013] input the optimal temperature control input parameters into corresponding temperature control modules to realize temperature control of the CT system.

[0014] On the basis of the above technical solution, the application can also be improved as follows:

[0015] Optionally, the determination of the number of hidden layer nodes based on the number of input layer nodes, the number of output layer nodes and the adjustment constant comprises:

[0016] the number of hidden layer nodes is calculated through formula (1);

[0017]

[0018] wherein h is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, and a is an adjustment constant between 1 and 10.

[0019] Optionally, the real-time collection of the temperature data to be adjusted of the X-ray detector and the input of the temperature data to be adjusted into the optimal temperature control model to obtain optimal temperature control input parameters comprise:

[0020] real-time collection of an inlet air temperature, a detector ambient temperature, a detector surface temperature, a CT system output temperature and a detector target temperature of a current X-ray detector temperature control system;

[0021] input of the inlet air temperature, the detector ambient temperature, the detector surface temperature, the CT system output temperature and the detector target temperature into the optimal temperature control model to obtain an inlet air fan rotating speed, an outlet air fan rotating speed, a detector water cooling flow rate and a heating belt target temperature.

[0022] Optionally, the input of the temperature data to be adjusted into the optimal temperature control model to obtain optimal temperature control input parameters comprises:

[0023] calculation of the optimal temperature control input parameters through formula (2);

[0024] y t =W y ·h t +b y formula (2);

[0025] In the formula, y t is the optimal temperature control input parameter, W y is the output layer weight matrix, h t is the hidden layer state at the current moment, b y is the output layer bias term.

[0026] Optionally, further comprising:

[0027] processing the time sequence correlation of the temperature data to be adjusted by a gating unit, the gating unit including a forget gate, an input gate and an output gate;

[0028] Based on the time sequence correlation, the temperature control model can accurately predict temperature changes and output the optimal temperature control input parameter.

[0029] Optionally, during the training of the temperature control model, the gradient descent method is used to continuously update the weights and biases of the temperature control model until the loss function value of the temperature control model converges to a preset threshold.

[0030] Optionally, the temperature data sample is input into the temperature control model for training to obtain the optimal temperature control model, further comprising:

[0031] calculating the loss function value of the temperature control model by formula (3);

[0032] L = (t'-t us ) 2 Formula (3);

[0033] In the formula, L is the loss function of the temperature control model, t' is the predicted value, t us is the output value;

[0034] optimizing the temperature control model based on the loss function value.

[0035] A CT system temperature control system based on a long short-term memory neural network, comprising:

[0036] A model acquisition module is configured to collect temperature data samples of an X-ray detector under different conditions and then input the temperature data samples into a pre-constructed temperature control model for training to obtain an optimal temperature control model.

[0037] The temperature control model includes an input layer, a hidden layer and an output layer, the input layer includes an air inlet temperature node, a detector environment temperature node, a detector surface temperature node, a CT system output temperature node and a detector target temperature node, the output layer includes an air inlet fan speed node, an air outlet fan speed node, a detector water cooling flow rate node and a heating belt target temperature node, and the number of hidden layer nodes is determined based on the number of input layer nodes, the number of output layer nodes and an adjustment constant.

[0038] a model deployment module configured to deploy the optimal temperature control model to an X-ray detector temperature control system;

[0039] a parameter acquisition module configured to acquire real-time temperature data of the X-ray detector to be adjusted, input the temperature data to be adjusted into the optimal temperature control model, and obtain optimal temperature control input parameters;

[0040] an execution module configured to input the optimal temperature control input parameters into a corresponding temperature control module to realize temperature control of the CT system.

[0041] An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the method when executing the computer program.

[0042] A non-transitory computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of the method.

[0043] The present application has the following advantages:

[0044] The CT system temperature control method based on the long short-term memory neural network in the present application captures the time sequence characteristics of temperature changes through the long short-term memory neural network, solves the hysteresis problem of traditional algorithms, accurately controls the temperature fluctuation of the X-ray detector within ±0.5℃, and guarantees the image quality. The multi-input and output node design can automatically learn the multi-variable coupling relationship of the inlet temperature, fan speed, and the like, adapt to complex working conditions such as continuous scanning and environmental fluctuations, and does not need manual calibration. The temperature control model forms an intelligent closed-loop control after training, balances the precision and efficiency, is easy to deploy in engineering, and effectively improves the stability and diagnosis reliability of the CT system. BRIEF DESCRIPTION OF DRAWINGS

[0045] For the purpose of illustration and not limitation, the present application will now be described in conjunction with embodiments thereof and the accompanying drawings, in which:

[0046] Figure 1 A flowchart of the CT system temperature control method based on the long short-term memory neural network in the embodiments of the present application;

[0047] Figure 2 A schematic diagram of the main components of the CT system temperature control system based on the long short-term memory neural network in the embodiments of the present application;

[0048] Figure 3 A schematic diagram of the neuron input and output interface in the embodiments of the present application;

[0049] Figure 4A schematic diagram of a gating unit in an embodiment of the present application;

[0050] Figure 5 A schematic diagram of a neural network structure in an embodiment of the present application;

[0051] Figure 6 A schematic diagram of an ELU curve in an embodiment of the present application;

[0052] Figure 7 A schematic diagram of an electronic device physical structure provided by the present application. DETAILED DESCRIPTION

[0053] In order to make the personnel in the technical field better understand the present application scheme, the technical scheme in the embodiment of the present application will be described clearly and completely in combination with the drawings in the embodiment of the present application. Obviously, the described embodiment is only a part of the embodiment of the present application, not all the embodiments. Based on the embodiment in the present application, all other embodiments obtained by the person skilled in the art without creative labor should belong to the scope of protection of the present application.

[0054] It should be noted that the terms "first", "second" and the like in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0055] It should be noted that the features in the embodiments of the present application and the embodiments can be combined with each other without conflict. The embodiments of the present application will be described in detail below in combination with the drawings.

[0056] Figure 1 A flowchart of a CT system temperature control method based on a long short-term memory neural network in an embodiment of the present application, as shown in Figure 1 The CT system temperature control method based on a long short-term memory neural network provided by the present application includes the following steps S101 to S104.

[0057] S101, collecting temperature data samples of the X-ray detector under different conditions and then inputting the pre-constructed temperature control model for training to obtain an optimal temperature control model.

[0058] It covers different scan modes of CT equipment (regular, spiral, etc.), load intensity (empty to continuous high load), machine room environment (room temperature of 18-26°C, humidity of 40-60%), and equipment new and old state, and also includes fan failure, temperature overrun and other abnormal working conditions. The inlet temperature, detector ambient temperature, surface temperature, CT system output temperature and target temperature are recorded synchronously by sensors at a sampling frequency of 10 Hz (once every 0.1 second), ensuring the capture of temperature dynamic changes. The collected data are time-stamped, outliers are removed, and missing values are completed to form a sample set containing multi-dimensional temperature parameters, providing comprehensive and accurate basic data for subsequent model training and ensuring the adaptability of the model to complex working conditions.

[0059] The temperature control model includes an input layer, a hidden layer and an output layer. The input layer includes an inlet temperature node, a detector ambient temperature node, a detector surface temperature node, a CT system output temperature node and a detector target temperature node. The output layer includes an inlet fan speed node, an outlet fan speed node, a detector water cooling flow rate node and a heating belt target temperature node. The number of hidden layer nodes is determined based on the number of input layer nodes, the number of output layer nodes and the adjustment constant.

[0060] The number of input layer nodes is determined by the current inlet temperature (t i ), the inlet temperature (t o ), the detector ambient temperature (t v ), the detector surface temperature (t c ) and the detector target temperature (t t ).

[0061] The number of output layer nodes is determined by the inlet fan speed (r i ), the outlet fan speed (r o ), the detector water cooling flow rate (w v ) and the heating belt target temperature (t b ).

[0062] The output parameters include the fan speeds r i and r o , the flow rate w v , and the temperature t b .

[0063] The number of hidden layer nodes is calculated by formula (1);

[0064]

[0065] In the formula, h is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, and a is an adjustment constant between 1 and 10.

[0066] Based on the number of input and output nodes, the input layer m of the Long Short-Term Memory (LSTM) neural network is determined to be 5, and the output layer n is determined to be 4. Following the principle that the number of hidden layer nodes should not exceed twice the number of input layer nodes, the adjustment constant is set to 9. Using the node number calculation formula, h≈12, and the final number of neurons in the hidden layer is selected as 12 (including a bias b). The number of hidden layer neurons in each layer is the same (i.e., 12 neurons).

[0067] A schematic diagram of the neural network structure of this invention is shown below. Figure 5 As shown, the number of hidden layers in a Long Short-Term Memory (LSTM) neural network is a crucial factor affecting its performance. Multiple hidden layers are used to fit nonlinear functions. Theoretically, the more hidden layers, the stronger the network's ability to map nonlinear data to the input. However, in practice, this also increases the network's complexity and training difficulty. It may also lead to overfitting (i.e., the model performs well on training data but poorly on new data). Furthermore, it may make the network more prone to getting trapped in local minima during the learning process, affecting the model's generalization ability. Similarly, fewer hidden layers may lead to underfitting, limiting the model's ability to capture complex patterns in the data, resulting in poor performance when faced with new data or complex tasks.

[0068] The CT radiation source detector temperature control system of this invention is a typical nonlinear mapping. The goal of this invention is to control the temperature of the CT radiation source detector within a specified target temperature range with high precision, based on a long short-term memory (LSM) neural network model. Considering the characteristics of the CT radiation source detector temperature control system and the input / output parameters of the LSM neural network, while also taking into account network performance and training complexity, the number of hidden layers is determined to be 10 layers.

[0069] Neuron input / output interfaces, such as Figure 3 As shown, where:

[0070] ct-1: Represents the memory state (value) at the previous moment;

[0071] ct: Represents the current memory state (value);

[0072] ht-1: Represents the hidden state (value) at the previous time step;

[0073] ht: Represents the hidden state (value) at the current moment;

[0074] xt: Represents the input parameter (value) at the current moment;

[0075] yt: Represents the output parameter (value) at the current time.

[0076] The internal algorithm of the neuron comprises three gating units (forgetting gate, input gate and output gate). Figure 4

[0077] it: represents the input gate; i t = sigma (W i [h t-1 , x t ])+b i

[0078] ot: represents the output gate; f t = sigma (W f [h t-1 , x t ])+b f

[0079] ft: represents the forgetting gate; o t = sigma (W o [h t-1 , x t ])+b o .

[0080] Common activation functions include Sigmod (S-shaped function), Softmax (flexible maximum transmission function), tanh function (hyperbolic tangent function), ReLU (rectified linear unit function), LeakyReLU (leaky linear unit function), ELU (exponential linear unit function) and the like. Among them, the Sigmod activation function maps real numbers to the interval (0, 1), and is commonly used to solve binary classification problems. The Softmax activation function normalizes real numbers to the interval (0, 1) and makes the sum equal to 1, and is commonly used to solve multi-classification problems. The tanh activation function maps real numbers to the interval (-1, 1), and is commonly used to solve multi-classification or regression problems. The ReLU activation function outputs a positive number itself for a positive number, and outputs 0 for a negative number, but the disadvantage is that negative values can cause the neuron to “die”. The LeakyReLU function solves the problem of neuron “death” of the ReLU function, but has the problem of poor convergence. The ELU activation function has an output distribution of zero mean, which can speed up the training speed. At the same time, the ELU activation function is unilaterally saturated, which can better converge.

[0081] The CT ray source detector temperature control system of the application precisely controls the target temperature, which is a typical regression task. Considering the training speed and better convergence, the application adopts the ELU activation function.

[0082] The ELU curve is as shown in Figure 6 ; and the ELU activation function expression is:

[0083] ​

[0084] where alpha is a predefined hyper-parameter to control the saturation of the output when x is negative.

[0085] The time sequence correlation of the temperature data to be adjusted is processed by a gating unit, and the gating unit includes a forget gate, an input gate and an output gate.

[0086] During the model training process, the weights and biases of the temperature control model are continuously iteratively updated by using the gradient descent method until the loss function value of the temperature control model converges to a preset threshold, so as to ensure the prediction and control accuracy of the temperature control model on temperature change.

[0087] In the long short-term memory neural network, the initialization of the weight is crucial to the training effect. Improper selection of the weight can cause the network to fail to be trained correctly or overfit the data. Common methods of weight initialization include uniform distribution, Gaussian distribution, Xavier initialization, He initialization and the like.

[0088] The present scheme selects the ELU method based on the activation function, and the He initialization method (also known as Kaiming initialization) with better performance is used for weight initialization.

[0089] The He initialization function expression is:

[0090]

[0091] where n is the number of nodes in the hidden layer.

[0092] The bias in the long short-term memory neural network is mainly used to adjust the threshold value of the activation function of each node in the neural network, so as to help the neural network better fit the training data. The bias mainly ensures better fitting, improves the ability to handle complex problems, and more importantly, improves the non-linear performance of the model and prevents the model from overfitting.

[0093] The bias of the present application is set to 1 (i.e. b = 1).

[0094] Error backpropagation is the most important process of the long short-term memory neural network in the model training stage. The error signal is propagated in the reverse direction from the output layer through the hidden layer to the input layer. Backpropagation uses the gradient descent method to update the weight and bias values in each layer of the neural network.

[0095] The weight and bias values are updated according to the propagated error signal, and the gradient descent method is used to update the weight and bias values in the network. The learning rate setting can control the step size of parameter update, and the initial value of the learning rate is set to 0.01 in the present system.

[0096] The update formula of the weight parameter is:

[0097]

[0098] (w ′ : updated weight; w: weight before update; η: learning rate; L: loss function)

[0099] The update formula of the bias parameter is:

[0100]

[0101] (b ′ : updated bias; b: bias before update; η: learning rate; L: loss function).

[0102] The loss function value of the temperature control model is calculated by formula (3);

[0103] L = (t-t us ) 2 Formula (3);

[0104] In the formula, L is the loss function of the temperature control model, t' is the predicted value, t us is the output value.

[0105] S102, deploy the optimal temperature control model to the X-ray detector temperature control system.

[0106] S103, based on the X-ray detector temperature control system, real-time acquisition of the X-ray detector to be adjusted temperature data, input the to-be-adjusted temperature data into the optimal temperature control model, to obtain the optimal temperature control input parameter.

[0107] The to-be-adjusted temperature data mainly involves the CT system input temperature, the detector environment temperature, the detector surface temperature, and the CT system output temperature. The CT system input temperature is the inlet temperature, which is affected by the temperature of the external environment. The detector environment temperature is the operating environment temperature of the X-ray source detector, which is affected by the superposition of the inlet temperature and the temperature generated by the detector. The detector environment temperature of the present application is the temperature of the water in the detector water cooling system. The detector surface temperature is the target temperature of the X-ray source detector, which is affected by the CT system input temperature, the detector environment temperature, and the CT system output temperature. The CT system output temperature is the outlet temperature, which is affected by the inlet temperature, the temperature generated by the CT system, the air flow rate, the heating system and many other factors.

[0108] The inlet air temperature, the detector environment temperature, the detector surface temperature, the CT system output temperature and the detector target temperature of the current X-ray detector temperature control system are collected in real time, and the inlet air temperature, the detector environment temperature, the detector surface temperature, the CT system output temperature and the detector target temperature are input into the optimal temperature control model to output the inlet air fan rotating speed, the outlet air fan rotating speed, the detector water cooling flow rate and the heating belt target temperature.

[0109] The optimal temperature control input parameter is calculated by formula (2);

[0110] y t =W y ·h t +b y Formula (2);

[0111] In the formula, y t is the optimal temperature control input parameter, W y is the output layer weight matrix, h t is the hidden layer state at the current moment, and b y is the output layer bias term.

[0112] S104, input the optimal temperature control input parameter into the corresponding temperature control module to realize the temperature control of the CT system.

[0113] Figure 2 is the schematic diagram of the main components of the CT system temperature control system based on the long short-term memory neural network in the embodiment of the application. As shown in Figure 2 , the CT system temperature control system based on the long short-term memory neural network provided in the embodiment of the application 1 includes a model acquisition module 10, a model deployment module 20, a parameter acquisition module 30 and an execution module 40.

[0114] The system hardware includes one Sinofer CT scanner and one AI server. The CT scanner provides an X-ray detector, and the AI server is used for training a model.

[0115] The model acquisition module 10 is used for collecting temperature data samples of the X-ray detector under different conditions and then inputting the temperature control model constructed in advance for training to obtain an optimal temperature control model.

[0116] The temperature control model 10 includes an input layer, a hidden layer and an output layer. The input layer includes an inlet air temperature node, a detector environment temperature node, a detector surface temperature node, a CT system output temperature node and a detector target temperature node. The output layer includes an inlet air fan rotating speed node, an outlet air fan rotating speed node, a detector water cooling flow rate node and a heating belt target temperature node. The number of hidden layer nodes is determined based on the number of input layer nodes, the number of output layer nodes and an adjustment constant.

[0117] a model deployment module 20, configured to deploy the optimal temperature control model to an X-ray detector temperature control system;

[0118] a parameter acquisition module 30, configured to acquire temperature data to be adjusted of the X-ray detector based on the X-ray detector temperature control system in real time, and input the temperature data to be adjusted into the optimal temperature control model to obtain optimal temperature control input parameters;

[0119] an execution module 40, configured to input the optimal temperature control input parameters into a corresponding temperature control module to implement temperature control of the CT system.

[0120] The CT system temperature control system based on the long short-term memory neural network is software arranged in a control system of a CT machine.

[0121] Figure 7 An electronic device entity structure schematic diagram provided by the embodiment of the present application is shown in the figure. Figure 7 The electronic device 50 includes a processor 501, a memory 502 and a bus 503.

[0122] The processor 501 and the memory 502 communicate with each other through the bus 503.

[0123] The processor 501 is configured to call program instructions in the memory 502 to execute the method provided by each method embodiment and execute the method provided by the embodiment of the present application.

[0124] The embodiment provides a non-transitory computer readable storage medium, which stores computer instructions, and the computer instructions make a computer execute the method provided by the embodiment of the present application.

[0125] Those skilled in the art can understand that all or part of the steps of the above method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program executes the steps of the above method embodiments when executed; and the foregoing storage medium includes ROM, RAM, magnetic disc or optical disc and various storage media that can store program codes.

[0126] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can occur depending on design requirements and other factors. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A CT system temperature control method based on a long short-term memory neural network, characterized by, The application comprises the following steps: Collecting temperature data samples of the X-ray detector under different conditions and inputting the pre-constructed temperature control model for training to obtain an optimal temperature control model; The temperature control model comprises an input layer, a hidden layer and an output layer, the input layer comprises an inlet air temperature node, a detector environment temperature node, a detector surface temperature node, a CT system output temperature node and a detector target temperature node, the output layer comprises an inlet air fan speed node, an outlet air fan speed node, a detector water cooling flow rate node and a heating belt target temperature node, and the number of hidden layer nodes is determined based on the number of input layer nodes, the number of output layer nodes and an adjustment constant; Deploying the optimal temperature control model to the X-ray detector temperature control system; Based on the X-ray detector temperature control system, collecting the temperature data to be adjusted of the X-ray detector, inputting the temperature data to be adjusted into the optimal temperature control model to obtain optimal temperature control input parameters; Inputting the optimal temperature control input parameters into the corresponding temperature control module to realize temperature control of the CT system. 2.The CT system temperature control method based on long short-term memory neural network of claim 1, wherein, The number of hidden layer nodes is determined based on the number of input layer nodes, the number of output layer nodes and an adjustment constant, comprising: Calculating the number of hidden layer nodes by formula (1); In the formula, h is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, and a is an adjustment constant between 1 and 10. 3.The CT system temperature control method based on long short-term memory neural network of claim 1, wherein, The real-time collection of the temperature data to be adjusted of the X-ray detector, the inputting of the temperature data to be adjusted into the optimal temperature control model to obtain optimal temperature control input parameters, comprising: Real-time collection of the inlet air temperature, the detector environment temperature, the detector surface temperature, the CT system output temperature and the detector target temperature of the current X-ray detector temperature control system; Inputting the inlet air temperature, the detector environment temperature, the detector surface temperature, the CT system output temperature and the detector target temperature into the optimal temperature control model to obtain the inlet air fan speed, the outlet air fan speed, the detector water cooling flow rate and the heating belt target temperature. 4.The CT system temperature control method based on long short-term memory neural network of claim 1, wherein, The inputting of the temperature data to be adjusted into the optimal temperature control model to obtain optimal temperature control input parameters, comprising: Calculating the optimal temperature control input parameters by formula (2); y t = W y · h t + b y Equation (2); where y t is the optimal temperature control input parameter, W y is the output layer weight matrix, h t is the hidden layer state at the current time, b y is the output layer bias term. 5.The CT system temperature control method based on long short-term memory neural network of claim 1, wherein, Further comprising: Processing the time sequence correlation of the temperature data to be adjusted by a gating unit, the gating unit comprising a forget gate, an input gate and an output gate; Based on the time sequence correlation, the temperature control model can accurately predict temperature changes and output optimal temperature control input parameters. 6.The CT system temperature control method based on long short-term memory neural network of claim 5, wherein, In the training process of the temperature control model, the gradient descent method is used to continuously update the weights and biases of the temperature control model until the loss function value of the temperature control model converges to a preset threshold.

7. The long short-term memory neural network-based CT system temperature control method of claim 6, wherein, The inputting of the temperature data samples into the temperature control model for training to obtain an optimal temperature control model, further comprising: Calculating the loss function value of the temperature control model by formula (3); L = (t' - t us ) 2 Equation (3); In the formula, L is a loss function of the temperature control model, t' is a predicted value, t is an output value us ; Optimizing the temperature control model based on the loss function value.

8. A temperature control system for a CT system based on a long short-term memory neural network, characterized in that, The application comprises the following steps: The model obtaining module is configured to collect temperature data samples of the X-ray detector under different conditions and input the temperature control model constructed in advance for training to obtain an optimal temperature control model. The temperature control model comprises an input layer, a hidden layer and an output layer, the input layer comprises an inlet air temperature node, a detector ambient temperature node, a detector surface temperature node, a CT system output temperature node and a detector target temperature node, the output layer comprises an inlet air fan rotating speed node, an outlet air fan rotating speed node, a detector water cooling flow rate node and a heating belt target temperature node, and the number of hidden layer nodes is determined based on the number of input layer nodes, the number of output layer nodes and an adjustment constant. The model deployment module is configured to deploy the optimal temperature control model to the X-ray detector temperature control system. The parameter obtaining module is configured to collect to-be-adjusted temperature data of the X-ray detector based on the X-ray detector temperature control system in real time, input the to-be-adjusted temperature data into the optimal temperature control model, and obtain optimal temperature control input parameters. The execution module is configured to input the optimal temperature control input parameters into a corresponding temperature control module to realize temperature control of the CT system.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, The computer program is executed by the processor to implement the steps of the method according to any one of claims 1 to 7.

10. A non-transitory computer readable medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method according to any one of claims 1 to 7.

Images