Research and development cost proportion automatic calculation method and system based on artificial intelligence
By collecting and processing historical R&D project data in power grid enterprises and building an improved neural network model, the subjectivity and inefficiency of traditional manual calculation methods were solved, and rapid and accurate R&D expense ratio prediction was achieved, thus optimizing resource allocation and management decisions.
Patent Information
- Application Number
- CN202510869248.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-26
- Publication Date
- 2025-09-26
AI Technical Summary
The traditional method of calculating the ratio of R&D expenses relies on manual experience and is easily influenced by personal subjectivity, causing the calculation results to deviate from actual needs and failing to fully utilize the company's historical data, affecting the accuracy and efficiency of decision-making.
By collecting historical R&D project data from the power grid industry, performing preprocessing, filling missing values, and detecting outliers, an improved neural network algorithm model is constructed, and random forest is used to screen key variables. A R&D expense ratio prediction model is established to achieve automated calculation.
Quickly and accurately predict the proportion of R&D expenses, reduce manpower and time costs, improve decision-making efficiency, optimize resource allocation, reduce operating costs, and increase project management flexibility.
Smart Images
Figure CN120705443A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of R&D expense forecasting, and in particular relates to an artificial intelligence-based method and system for automatically calculating the ratio of R&D expenses. Background Art
[0002] With the widespread adoption of smart grids and new energy technologies, and strong policy support for scientific and technological innovation, power grid companies face pressure and challenges to increase R&D investment, optimize resource allocation, and enhance innovation capabilities. As a driving force for technological innovation and product upgrades, R&D project funding is a key factor in a company's sustainable development and competitiveness. However, how to rationally allocate R&D expenses to improve R&D efficiency and the conversion rate of research results, thereby meeting the strategic planning and decision-making support needs of power grid companies, has always been a challenge. Traditional methods for calculating R&D expense ratios often rely on manual experience. This approach has the following shortcomings: manual calculation of R&D expense ratios is easily influenced by personal experience and subjective will, which may cause the calculated results to deviate from actual needs and affect the accuracy of R&D decisions. Furthermore, manual calculation of R&D expense ratios often involves tedious data collection and manual analysis, which is time-consuming and labor-intensive. Moreover, traditional calculation methods often fail to fully utilize the extensive historical R&D project data accumulated by companies, thus failing to fully realize the value of this data.
[0003] In order to overcome these shortcomings, a more scientific, efficient and objective method for calculating the R&D expense ratio is urgently needed to improve the level of R&D management and decision-making quality of power companies. Summary of the Invention
[0004] In order to solve the above problems existing in the prior art, the present invention proposes an automatic calculation method and system for the R&D expense ratio based on artificial intelligence.
[0005] The purpose of the present invention can be achieved through the following technical solutions: An artificial intelligence-based automatic calculation method for the ratio of R&D expenses, comprising: Collect historical R&D project data, including but not limited to historical project type, historical project scale, historical project cycle, historical technical difficulty, historical R&D expenses, historical R&D team size, and historical R&D expense ratio; A research and development project dataset is obtained by preprocessing the historical research and development project data; Constructing an R&D expense ratio prediction model based on the R&D project data set by using an improved neural network algorithm; Obtain current R&D project data, including but not limited to project type, project scale, project cycle, technical difficulty, R&D cost, and R&D team size, and obtain the current R&D cost ratio based on the R&D project data using the R&D cost ratio prediction model.
[0006] Preferably, the data collection scope of the historical R&D project data is R&D projects of different types, scales, and cycles in the power grid industry, and the power grid industry includes but is not limited to the power transmission field, power transformation field, power distribution field, new energy field, and smart grid field; The data sources of the historical R&D project data include but are not limited to the internal database of the power grid company, power grid industry reports, and public power grid R&D project information; The project types of the historical R&D project data include but are not limited to technological innovation, product development, technological transformation, and experimental research. The project scales of the historical R&D project data include small scale, medium scale, and large scale. The small scale refers to the R&D of a single device or component, the medium scale refers to the R&D of a subsystem or module, and the large scale refers to the R&D of the entire power grid system or key technologies.
[0007] Preferably, constructing the R&D project dataset by preprocessing the historical R&D project data includes: Obtaining historical R&D project standard data through preprocessing of the historical R&D project data, wherein the preprocessing includes missing value filling, outlier detection, and data standardization; According to the historical R&D project standard data, the historical R&D expense ratio is used as the target variable, and the remaining historical R&D project standard data are used as feature variables to construct a corresponding R&D project data set.
[0008] Preferably, the missing value filling is performed by a multiple interpolation filling method, and the mice package of the R language statistical software package is used to perform multiple interpolation; the outlier detection is performed by an isolation forest algorithm to detect outliers and delete them; the data standardization is performed by a Z-score standardization method to scale the data to eliminate the influence of different dimensions.
[0009] Preferably, the Isolation Forest algorithm includes: When building each tree, a feature is randomly selected; On the selected feature, randomly select a split value to split the data; Repeat the above steps to build the isolation tree until the given path length has reached the preset threshold, the data point has been isolated, or only one data point remains; Repeatedly constructing multiple isolation trees to form an isolation forest, where each tree in the isolation forest is based on a different random feature and a random split value; Preset outlier threshold; Obtaining outliers through anomaly score detection according to the isolation forest; When the anomaly score of a data point exceeds the preset anomaly threshold, the data point is abnormal.
[0010] Preferably, the improved neural network algorithm includes: Initialize the random forest regressor; Setting random forest parameters, including but not limited to the number of trees, tree depth, and minimum number of samples for leaf nodes; Training a random forest model based on the R&D project training set and obtaining important characteristic factors of the R&D project through key variable screening; Inputting the important characteristic factors of the R&D project into a neural network for training to obtain an initial model for predicting the R&D expense ratio; The R&D expense ratio prediction model is obtained by evaluating the R&D expense ratio initial model using evaluation indicators according to the R&D project test set.
[0011] Preferably, the key variable screening includes: Preset scoring thresholds for important features of R&D projects; Calculating the mean square error of the R&D project training set without feature segmentation; For each R&D project characteristic variable, try all possible split points and calculate the mean square error between the two subsets after splitting; For each split point of each R&D project characteristic variable, calculate the reduction in mean square error before and after the split; For each R&D project characteristic variable, the mean square error reduction of all possible split points is averaged to obtain the importance score of the R&D project characteristic variable; The R&D project characteristic variable whose importance score is higher than the R&D project important characteristic score threshold is selected as the important characteristic factor of the R&D project.
[0012] An artificial intelligence-based automatic calculation system for R&D expense ratio, applied to the above-mentioned artificial intelligence-based automatic calculation method for R&D expense ratio, comprising a data collection and processing module, a model building module, an expense ratio prediction module, and a user interaction module; The data collection and processing module is used to collect historical R&D project data, and perform missing value filling, outlier detection and data standardization on the collected historical R&D project data, so as to construct an R&D project data set suitable for artificial intelligence model training; The model building module is used to build an R&D expense ratio prediction model based on the R&D project data set by using an improved neural network algorithm; The expense ratio prediction module is used to obtain data related to the current R&D project and predict the R&D expense ratio of the current R&D project through the R&D expense ratio prediction model; The user interaction module is used to provide a user interface, allowing the user to input data related to the current R&D project, start the R&D expense ratio forecasting process, and display the final R&D expense ratio forecasting results.
[0013] An electronic device includes a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, the above-mentioned method for automatically calculating the ratio of R&D expenses based on artificial intelligence is implemented.
[0014] A storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the above-mentioned artificial intelligence-based automatic calculation method for the ratio of R&D expenses.
[0015] The beneficial effects of the present invention are: By collecting historical R&D project data and constructing a high-quality R&D project data set through preprocessing, and constructing an R&D expense ratio prediction model based on the R&D project data set through an improved neural network algorithm, it is possible to identify complex patterns of R&D expense expenditures based on the analysis of historical R&D project data, predict the R&D expense ratio more quickly and accurately, reduce the inconsistency of calculation results caused by differences in personal subjective judgment, and save manpower and time.
[0016] By obtaining current R&D project data and using the R&D expense ratio prediction model to obtain the current R&D expense ratio based on the R&D project data, the R&D expense ratio can be quickly provided using an automated calculation method, helping management and decision makers to quickly make decisions on R&D resource allocation, improve decision-making efficiency, and optimize resource allocation.
[0017] By training the random forest model to screen key variables, we obtained the important characteristic factors of R&D projects and input them into the neural network training and testing to obtain the R&D expense ratio prediction model, which effectively improved the model's prediction performance, reduced computing costs and overfitting risks, and enhanced the model's robustness.
[0018] By accurately predicting R&D expenses, power grid companies can better control costs, avoid unnecessary expenses, and thus reduce overall operating costs; by real-time monitoring and adjusting the progress of R&D projects and adjusting the cost ratio forecast according to actual conditions, the flexibility of project management can be improved; and by improving the neural network model, it can be continuously adjusted and optimized according to new data to adapt to the ever-changing market and technological environment. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] To facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.
[0020] Figure 1 The figure is a flow chart of an automatic calculation method of the R&D expense ratio based on artificial intelligence of the present invention. DETAILED DESCRIPTION
[0021] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.
[0022] See also Figure 1 , an automatic calculation method for R&D expense ratio based on artificial intelligence, including: S1: Collect historical R&D project data, including but not limited to historical project type, historical project scale, historical project cycle, historical technical difficulty, historical R&D expenses, historical R&D team size, and historical R&D expense ratio; S2: constructing a research and development project dataset based on the historical research and development project data through preprocessing, wherein the preprocessing includes missing value filling, outlier detection and data standardization; S3: Constructing an R&D expense ratio prediction model based on the R&D project data set by using an improved neural network algorithm; S4: Obtain current R&D project data, including but not limited to project type, project scale, project cycle, technical difficulty, R&D cost, and R&D team size, and obtain the current R&D cost ratio based on the R&D project data using the R&D cost ratio prediction model.
[0023] In this embodiment, the collection of historical R&D project data specifically includes: The data collection scope of the historical R&D project data is R&D projects of different types, scales, and cycles in the power grid industry, including but not limited to the power transmission, substation, distribution, new energy, and smart grid fields; The data sources of the historical R&D project data include but are not limited to the internal database of the power grid company, power grid industry reports, and public power grid R&D project information; The project types of the historical R&D project data include but are not limited to technological innovation, product development, technological transformation, and experimental research. The project scales of the historical R&D project data include small scale, medium scale, and large scale. Specifically, the small scale is the R&D of a single device or component, the medium scale is the R&D of a subsystem or module, and the large scale is the R&D of the entire power grid system or key technologies.
[0024] In this embodiment, the R&D project dataset is constructed by preprocessing the historical R&D project data through the following steps: S201: Obtaining historical R&D project standard data through preprocessing according to the historical R&D project data; S201-1: The missing value filling method processes missing values through a multiple imputation filling method. The multiple imputation filling method does not rely on any single model, but combines the predictions of multiple models, thereby reducing bias and uncertainty in the imputation process; In this embodiment, multiple imputation is performed using the mice package, a statistical software package in the R language, so that an imputation model suitable for the data type is automatically selected and executed.
[0025] S201-2: The outlier detection uses the Isolation Forest algorithm to detect outliers and delete them; The Isolation Forest algorithm includes: S201-21: When building each tree, a feature is randomly selected; S201-22: Randomly select a split value on the selected feature to split the data; S201-23: Repeat the above steps to build an isolation tree until the given path length has reached a preset threshold, the data point has been isolated (cannot be further split), or only one data point remains; S201-24: repeatedly constructing multiple isolation trees to form an isolation forest, where each tree in the isolation forest is based on a different random feature and a random split value; S201-25: preset outlier threshold; S201-26: Obtaining an outlier by performing anomaly score detection according to the isolation forest; Specifically, for each data point, the average path length in all isolation trees is calculated. The shorter the path length, the greater the possibility that the data point is an outlier. The mathematical expression of the anomaly score is: , Where S(x) is the anomaly score of data point x, length(x) is the path length of data point x in the isolation tree, and lenV is the average path length of leaf nodes in all isolation trees; S201-27: When the anomaly score of a data point exceeds the preset anomaly value threshold, the data point is abnormal; S201-3: The data normalization is performed by scaling the data using a Z-score normalization method to eliminate the effects of different dimensions; S202: Constructing a research and development project dataset based on the historical research and development project standard data.
[0026] Specifically, the historical R&D expense ratio is used as the target variable, and the standard data of other historical R&D projects are used as feature variables to construct the corresponding R&D project data set.
[0027] It should be noted that in order to merge data collected from different sources into a unified dataset, it is necessary to identify and delete duplicate records, and convert text-type data such as project type and project scale into numerical form through label encoding. For numerical data, the above-mentioned preprocessing operations are performed.
[0028] In this embodiment, the R&D expense ratio prediction model is constructed based on the R&D project data set by using an improved neural network algorithm, specifically by the following steps: S301: obtaining an R&D project training set and an R&D project test set by dividing the R&D project data set; S302: Obtaining an initial model for predicting the R&D expense ratio by training an improved neural network algorithm based on the R&D project training set; The improved neural network algorithm is specifically implemented by the following steps: S302-1: Initialize random forest regressor; S302-2: Setting random forest parameters, including but not limited to the number of trees, tree depth, and minimum number of leaf node samples; S302-3: Training a random forest model based on the R&D project training set and obtaining important characteristic factors of the R&D project through key variable screening; The key variable screening includes: S302-31: Preset scoring thresholds for important features of R&D projects; S302-32: Calculating the mean square error of the R&D project training set when feature segmentation is not performed; S302-33: For each R&D project characteristic variable, try all possible split points and calculate the mean square error of the two subsets after splitting; S302-34: For each split point of each R&D project characteristic variable, calculate the reduction in mean square error before and after the split; S302-35: For each R&D project characteristic variable, average the mean square error reduction of all possible split points to obtain the importance score of the R&D project characteristic variable; S302-36: Select the R&D project characteristic variable whose importance score is higher than the R&D project important characteristic score threshold as the R&D project important characteristic factor.
[0029] S302-4: Inputting the important characteristic factors of the R&D project into a neural network for training to obtain an initial model for predicting the R&D expense ratio; It should be noted that during the neural network training process, the neural network hyperparameters are optimized through the grid search algorithm, and the training is iterated until the preset maximum number of iterations is met.
[0030] S303: Evaluate the initial R&D expense ratio prediction model using evaluation indicators based on the R&D project test set to obtain an R&D expense ratio prediction model; The evaluation metrics include mean square error, mean absolute error, and R² score.
[0031] Specifically, based on the results of the evaluation indicators, analyze whether the initial model for predicting the R&D expense ratio meets the preset expectations; when the mean square error and the mean absolute error are both higher than the preset error threshold, it means that the initial model for predicting the R&D expense ratio has a large prediction error at certain points; when the R² score is lower than the preset R² score threshold, it means that the model does not capture the trend of the R&D project data well, that is, the initial model for predicting the R&D expense ratio needs to be adjusted, such as adjusting the model structure or continuing to optimize the hyperparameters.
[0032] When the mean square error and the mean absolute error are both lower than the preset error threshold and the R² score is higher than the preset R² score threshold, the current R&D expense ratio prediction initial model is the R&D expense ratio prediction model.
[0033] In this embodiment, a R&D expense ratio prediction model is obtained by training and testing the R&D project data set, and the complex mapping relationship between the R&D expense ratio and various features is learned through a nonlinear model, in order to predict a reasonable R&D expense ratio under new conditions.
[0034] An artificial intelligence-based automatic calculation system for R&D expense ratio, including a data collection and processing module, a model building module, an expense ratio prediction module, and a user interaction module; The data collection and processing module is used to collect historical R&D project data, and perform missing value filling, outlier detection and data standardization on the collected historical R&D project data, so as to construct an R&D project data set suitable for artificial intelligence model training; The model building module is used to build an R&D expense ratio prediction model based on the R&D project data set by using an improved neural network algorithm; The expense ratio prediction module is used to obtain data related to the current R&D project and predict the R&D expense ratio of the current R&D project through the R&D expense ratio prediction model; The user interaction module is used to provide a user interface, allowing the user to input data related to the current R&D project, start the R&D expense ratio forecasting process, and display the final R&D expense ratio forecasting results.
[0035] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.
[0036] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0037] The program code included in the computer-readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF or the like, or any suitable combination thereof. The computer program code for performing the operation of the present invention can be written in one or more programming languages or a combination thereof, and the programming language includes an object-oriented programming language such as Java, Smalltalk, C++, and also includes a conventional procedural programming language such as "C" language or similar programming language. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, utilizing an Internet service provider to connect through the Internet).
[0038] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.
Claims
1. An automatic calculation method for the ratio of R&D expenses based on artificial intelligence, characterized in that: include: Collect historical R&D project data, including but not limited to historical project type, historical project scale, historical project cycle, historical technical difficulty, historical R&D expenses, historical R&D team size, and historical R&D expense ratio; A research and development project dataset is obtained by preprocessing the historical research and development project data; Constructing an R&D expense ratio prediction model based on the R&D project data set by using an improved neural network algorithm; Obtain current R&D project data, including but not limited to project type, project scale, project cycle, technical difficulty, R&D cost, and R&D team size, and obtain the current R&D cost ratio based on the R&D project data using the R&D cost ratio prediction model.
2. The method for automatically calculating the ratio of R&D expenses based on artificial intelligence according to claim 1, characterized in that: The data collection scope of the historical R&D project data is R&D projects of different types, scales, and cycles in the power grid industry, including but not limited to the power transmission, substation, distribution, new energy, and smart grid fields; The data sources of the historical R&D project data include but are not limited to the internal database of the power grid company, power grid industry reports, and public power grid R&D project information; The project types of the historical R&D project data include but are not limited to technological innovation, product development, technological transformation, and experimental research. The project scales of the historical R&D project data include small scale, medium scale, and large scale. The small scale refers to the R&D of a single device or component, the medium scale refers to the R&D of a subsystem or module, and the large scale refers to the R&D of the entire power grid system or key technologies.
3. The method for automatically calculating the ratio of R&D expenses based on artificial intelligence according to claim 1, characterized in that: The R&D project dataset constructed by preprocessing the historical R&D project data includes: Obtaining historical R&D project standard data through preprocessing of the historical R&D project data, wherein the preprocessing includes missing value filling, outlier detection, and data standardization; According to the historical R&D project standard data, the historical R&D expense ratio is used as the target variable, and the remaining historical R&D project standard data are used as feature variables to construct a corresponding R&D project data set.
4. The method for automatically calculating the ratio of R&D expenses based on artificial intelligence according to claim 3, characterized in that: The missing value filling method processes missing values by using the multiple imputation filling method, and the mice package of the R language statistical software package is used to perform multiple imputation; the outlier detection method detects outliers by using the Isolation Forest algorithm and deletes them; the data standardization method scales the data by using the Z-score standardization method to eliminate the influence of different dimensions.
5. The method for automatically calculating the ratio of R&D expenses based on artificial intelligence according to claim 4, characterized in that: The Isolation Forest algorithm includes: When building each tree, a feature is randomly selected; On the selected feature, randomly select a split value to split the data; Repeat the above steps to build the isolation tree until the given path length has reached the preset threshold, the data point has been isolated, or only one data point remains; Repeatedly constructing multiple isolation trees to form an isolation forest, where each tree in the isolation forest is based on a different random feature and a random split value; Preset outlier threshold; Obtaining outliers through anomaly score detection according to the isolation forest; When the anomaly score of a data point exceeds the preset anomaly threshold, the data point is abnormal.
6. The method for automatically calculating the ratio of R&D expenses based on artificial intelligence according to claim 1, characterized in that: The improved neural network algorithm includes: Initialize the random forest regressor; Setting random forest parameters, including but not limited to the number of trees, tree depth, and minimum number of samples for leaf nodes; Training a random forest model based on the R&D project training set and obtaining important characteristic factors of the R&D project through key variable screening; Inputting the important characteristic factors of the R&D project into a neural network for training to obtain an initial model for predicting the R&D expense ratio; The R&D expense ratio prediction model is obtained by evaluating the R&D expense ratio initial model using evaluation indicators according to the R&D project test set.
7. The method for automatically calculating the ratio of R&D expenses based on artificial intelligence according to claim 6, characterized in that: The key variable screening includes: Preset scoring thresholds for important features of R&D projects; Calculating the mean square error of the R&D project training set without feature segmentation; For each R&D project characteristic variable, try all possible split points and calculate the mean square error between the two subsets after splitting; For each split point of each R&D project characteristic variable, calculate the reduction in mean square error before and after the split; For each R&D project characteristic variable, the mean square error reduction of all possible split points is averaged to obtain the importance score of the R&D project characteristic variable; The R&D project characteristic variable whose importance score is higher than the R&D project important characteristic score threshold is selected as the important characteristic factor of the R&D project.
8. An artificial intelligence-based automatic calculation system for R&D expense ratio, applied to the artificial intelligence-based automatic calculation method for R&D expense ratio according to claims 1-7, characterized in that: It includes data collection and processing module, model building module, cost ratio prediction module, and user interaction module; The data collection and processing module is used to collect historical R&D project data, and perform missing value filling, outlier detection and data standardization on the collected historical R&D project data, so as to construct an R&D project data set suitable for artificial intelligence model training; The model building module is used to build an R&D expense ratio prediction model based on the R&D project data set through an improved neural network algorithm; The expense ratio prediction module is used to obtain data related to the current R&D project and predict the R&D expense ratio of the current R&D project through the R&D expense ratio prediction model; The user interaction module is used to provide a user interface, allowing the user to input data related to the current R&D project, start the R&D expense ratio forecasting process, and display the final R&D expense ratio forecasting results.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, it implements the method for automatically calculating the ratio of R&D expenses based on artificial intelligence as described in any one of claims 1 to 7.
10. A storage medium containing computer-executable instructions, characterized in that: When executed by a computer processor, the computer executable instructions are used to execute the artificial intelligence-based automatic calculation method for the R&D expense ratio as described in any one of claims 1 to 7.